diff --git a/cfa/catalog/public/datasets/stf/comprehensive_nssp_gold.toml b/cfa/catalog/public/datasets/stf/comprehensive_nssp_gold.toml deleted file mode 100644 index 94c8f91..0000000 --- a/cfa/catalog/public/datasets/stf/comprehensive_nssp_gold.toml +++ /dev/null @@ -1,46 +0,0 @@ -# This is the nssp latest comprehensive dataset. - -############################################################################ - -[properties] -# The name field is the only required field here. Though other -# files can be accessed in the datacat by key/name. - -name = "comprehensive_nssp_gold" -type = "reference" - -############################################################################ - -[source] - -storage_location.account = "cfaazurebatchprd" -storage_location.container = "nssp-etl" - -############################################################################ - -[prior_load] -# These are the details of where to store the reference data. -# If not included or given an empty string, the defaults in the -# catalog_defaults.toml will be used - -account = "" # use default -container = "" # use default -prefix = "dataops/stf/transformed/comprehensive_nssp_gold/" - -############################################################################ - -[load] -# These are the details of where to store the reference data. -# If not included or given an empty string, the defaults in the -# catalog_defaults.toml will be used - -account = "cfadagster" # explicit target account -container = "cfadagster" # explicit target container -prefix = "dagster-files/prod/comprehensive_nssp_gold/" - -############################################################################ - -# Changelog -# --------- -# 2026-03-04: Initial version created by Ryan Raasch (xng3) -# 2026-03-09: Changed name to comprehensive_nssp_gold diff --git a/cfa/catalog/public/datasets/stf/nssp_gold_v1.toml b/cfa/catalog/public/datasets/stf/nssp_gold_v1.toml index dc2b9e1..fae5020 100644 --- a/cfa/catalog/public/datasets/stf/nssp_gold_v1.toml +++ b/cfa/catalog/public/datasets/stf/nssp_gold_v1.toml @@ -12,8 +12,6 @@ name = "nssp_gold_v1" type = "experiment" -############################################################################ - [source] storage_location.account = "cfaazurebatchprd" diff --git a/cfa/catalog/public/datasets/stf/param_estimates.toml b/cfa/catalog/public/datasets/stf/param_estimates.toml index ea543f9..0409f54 100644 --- a/cfa/catalog/public/datasets/stf/param_estimates.toml +++ b/cfa/catalog/public/datasets/stf/param_estimates.toml @@ -9,6 +9,7 @@ name = "param_estimates" type = "reference" +schemas = "datasets/stf/schemas/param_estimates.py" ############################################################################ diff --git a/cfa/catalog/public/datasets/stf/schemas/nhsn_hrd.py b/cfa/catalog/public/datasets/stf/schemas/nhsn_hrd.py new file mode 100644 index 0000000..f9e58e4 --- /dev/null +++ b/cfa/catalog/public/datasets/stf/schemas/nhsn_hrd.py @@ -0,0 +1,3433 @@ +import random + +import pandas as pd +import pandera.pandas as pa +import polars as pl +from faker import Faker + +fake = Faker() +region_opts = [ + "AL", + "AR", + "AS", + "AZ", + "CA", + "CO", + "CT", + "DC", + "DE", + "FL", + "GU", + "HI", + "IL", + "IN", + "KS", + "ME", + "MN", + "MP", + "MS", + "MT", + "NE", + "NH", + "OH", + "OK", + "PA", + "PR", + "RI", + "Region 10", + "Region 2", + "Region 4", + "Region 6", + "Region 7", + "SC", + "SD", + "TN", + "USA", + "WI", + "ID", + "LA", + "MD", + "NJ", + "NM", + "OR", + "Region 8", + "VT", + "WA", + "AK", + "GA", + "MO", + "NV", + "Region 5", + "Region 9", + "UT", + "KY", + "MA", + "MI", + "NC", + "Region 1", + "TX", + "WY", + "Region 3", + "VA", + "VI", + "WV", + "NY", + "IA", + "ND", +] + +extract_schema = pa.DataFrameSchema( + { + "weekendingdate": pa.Column(object, coerce=True), + "jurisdiction": pa.Column( + str, pa.Check.isin(region_opts), nullable=True, coerce=True + ), + "numinptbeds": pa.Column(float, nullable=True, coerce=True), + "numinptbedsadult": pa.Column(float, nullable=True, coerce=True), + "numinptbedsped": pa.Column(float, nullable=True, coerce=True), + "numinptbedsocc": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccadult": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccped": pa.Column(float, nullable=True, coerce=True), + "numicubeds": pa.Column(float, nullable=True, coerce=True), + "numicubedsadult": pa.Column(float, nullable=True, coerce=True), + "numicubedsped": pa.Column(float, nullable=True, coerce=True), + "numicubedsocc": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccadult": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccped": pa.Column(float, nullable=True, coerce=True), + "numconfc19hosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19hosppatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfc19hosppats": pa.Column(float, nullable=True, coerce=True), + "totalconffluhosppats": pa.Column(float, nullable=True, coerce=True), + "numconfc19icupatsadult": pa.Column(float, nullable=True, coerce=True), + "totalconfc19icupats": pa.Column(float, nullable=True, coerce=True), + "totalconffluicupats": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadmped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmunk": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadm": pa.Column(float, nullable=True, coerce=True), + "totalconfflunewadm": pa.Column(float, nullable=True, coerce=True), + "pctinptbedsocc": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbeds": pa.Column(float, nullable=True, coerce=True), + "pctconffluinptbeds": pa.Column(float, nullable=True, coerce=True), + "pcticubedsocc": pa.Column(float, nullable=True, coerce=True), + "pctconfc19icubeds": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubeds": pa.Column(float, nullable=True, coerce=True), + "pctconfc19newadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfc19newadmped": pa.Column(float, nullable=True, coerce=True), + "numinptbedshosprep": pa.Column(float, nullable=True, coerce=True), + "numinptbedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedshosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "totalconfc19hosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "pcticubedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsperchosprep": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsperchosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19hosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsoccperchospre": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsoccperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19hosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccperchospre": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadult": pa.Column(float, nullable=True, coerce=True), + "pctinptbedsoccped": pa.Column(float, nullable=True, coerce=True), + "pcticubedsoccadult": pa.Column(float, nullable=True, coerce=True), + "pcticubedsoccped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19icubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicu": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicu": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "respseason": pa.Column(str, nullable=True, coerce=True), + "totalconfc19newadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfnewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19icupatsped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconffluhosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconffluhosppatsped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvhosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvhosppatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvhosppats": pa.Column(float, nullable=True, coerce=True), + "numconffluicupatsadult": pa.Column(float, nullable=True, coerce=True), + "numconffluicupatsped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvicupatsadult": pa.Column(float, nullable=True, coerce=True), + "numconfrsvicupatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvicupats": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfflunewadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmunk": pa.Column(float, nullable=True, coerce=True), + "numconfrsvnewadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvnewadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmunk": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvnewadm": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvinptbeds": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubeds": pa.Column(float, nullable=True, coerce=True), + "pctconfflunewadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfflunewadmped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvnewadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvnewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvinptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconffluhosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicu": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvhosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + } +) + +load_schema = pa.DataFrameSchema( + { + "weekendingdate": pa.Column(str, nullable=True, coerce=True), + "jurisdiction": pa.Column( + str, pa.Check.isin(region_opts), nullable=True, coerce=True + ), + "numinptbeds": pa.Column(float, nullable=True, coerce=True), + "numinptbedsadult": pa.Column(float, nullable=True, coerce=True), + "numinptbedsped": pa.Column(float, nullable=True, coerce=True), + "numinptbedsocc": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccadult": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccped": pa.Column(float, nullable=True, coerce=True), + "numicubeds": pa.Column(float, nullable=True, coerce=True), + "numicubedsadult": pa.Column(float, nullable=True, coerce=True), + "numicubedsped": pa.Column(float, nullable=True, coerce=True), + "numicubedsocc": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccadult": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccped": pa.Column(float, nullable=True, coerce=True), + "numconfc19hosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19hosppatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfc19hosppats": pa.Column(float, nullable=True, coerce=True), + "totalconffluhosppats": pa.Column(float, nullable=True, coerce=True), + "numconfc19icupatsadult": pa.Column(float, nullable=True, coerce=True), + "totalconfc19icupats": pa.Column(float, nullable=True, coerce=True), + "totalconffluicupats": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadmped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmunk": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadm": pa.Column(float, nullable=True, coerce=True), + "totalconfflunewadm": pa.Column(float, nullable=True, coerce=True), + "pctinptbedsocc": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbeds": pa.Column(float, nullable=True, coerce=True), + "pctconffluinptbeds": pa.Column(float, nullable=True, coerce=True), + "pcticubedsocc": pa.Column(float, nullable=True, coerce=True), + "pctconfc19icubeds": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubeds": pa.Column(float, nullable=True, coerce=True), + "pctconfc19newadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfc19newadmped": pa.Column(float, nullable=True, coerce=True), + "numinptbedshosprep": pa.Column(float, nullable=True, coerce=True), + "numinptbedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedshosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "totalconfc19hosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "pcticubedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsperchosprep": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsperchosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19hosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsoccperchospre": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsoccperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19hosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccperchospre": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadult": pa.Column(float, nullable=True, coerce=True), + "pctinptbedsoccped": pa.Column(float, nullable=True, coerce=True), + "pcticubedsoccadult": pa.Column(float, nullable=True, coerce=True), + "pcticubedsoccped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19icubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicu": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicu": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "respseason": pa.Column(str, nullable=True, coerce=True), + "totalconfc19newadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfnewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19icupatsped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconffluhosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconffluhosppatsped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvhosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvhosppatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvhosppats": pa.Column(float, nullable=True, coerce=True), + "numconffluicupatsadult": pa.Column(float, nullable=True, coerce=True), + "numconffluicupatsped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvicupatsadult": pa.Column(float, nullable=True, coerce=True), + "numconfrsvicupatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvicupats": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfflunewadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmunk": pa.Column(float, nullable=True, coerce=True), + "numconfrsvnewadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvnewadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmunk": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvnewadm": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvinptbeds": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubeds": pa.Column(float, nullable=True, coerce=True), + "pctconfflunewadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfflunewadmped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvnewadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvnewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvinptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconffluhosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicu": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvhosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + } +) + + +def extract_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: + data = { + "weekendingdate": [fake.date() for _ in range(size)], + "jurisdiction": [random.choice(region_opts) for _ in range(size)], + "numinptbeds": [random.uniform(7, 831870) for _ in range(size)], + "numinptbedsadult": [random.uniform(3, 743870) for _ in range(size)], + "numinptbedsped": [random.uniform(0, 84581) for _ in range(size)], + "numinptbedsocc": [random.uniform(0, 587461) for _ in range(size)], + "numinptbedsoccadult": [ + random.uniform(0, 543870) for _ in range(size) + ], + "numinptbedsoccped": [random.uniform(0, 84581) for _ in range(size)], + "numicubeds": [random.uniform(0, 100000) for _ in range(size)], + "numicubedsadult": [random.uniform(0, 90000) for _ in range(size)], + "numicubedsped": [random.uniform(0, 10000) for _ in range(size)], + "numicubedsocc": [random.uniform(0, 90000) for _ in range(size)], + "numicubedsoccadult": [random.uniform(0, 80000) for _ in range(size)], + "numicubedsoccped": [random.uniform(0, 10000) for _ in range(size)], + "numconfc19hosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfc19hosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19hosppats": [ + random.uniform(0, 60000) for _ in range(size) + ], + "totalconffluhosppats": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfc19icupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19icupats": [random.uniform(0, 15000) for _ in range(size)], + "totalconffluicupats": [random.uniform(0, 12000) for _ in range(size)], + "totalconfc19newadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult18to49": [ + random.uniform(0, 20000) for _ in range(size) + ], + "totalconfc19newadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfc19newadmunk": [random.uniform(0, 5000) for _ in range(size)], + "totalconfc19newadm": [random.uniform(0, 40000) for _ in range(size)], + "totalconfflunewadm": [random.uniform(0, 35000) for _ in range(size)], + "pctinptbedsocc": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbeds": [random.uniform(0, 100) for _ in range(size)], + "pctconffluinptbeds": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsocc": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19icubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconffluicubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19newadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19newadmped": [random.uniform(0, 100) for _ in range(size)], + "numinptbedshosprep": [random.uniform(0, 500000) for _ in range(size)], + "numinptbedsocchosprep": [ + random.uniform(0, 400000) for _ in range(size) + ], + "numicubedshosprep": [random.uniform(0, 100000) for _ in range(size)], + "numicubedsocchosprep": [ + random.uniform(0, 90000) for _ in range(size) + ], + "totalconfc19hosppatshosprep": [ + random.uniform(0, 60000) for _ in range(size) + ], + "totalconffluhosppatshosprep": [ + random.uniform(0, 50000) for _ in range(size) + ], + "totalconfrsvhosppatshosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfc19icupatshosprep": [ + random.uniform(0, 15000) for _ in range(size) + ], + "totalconffluicupatshosprep": [ + random.uniform(0, 12000) for _ in range(size) + ], + "totalconfrsvicupatshosprep": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19newadmpedhosprep": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19newadmadulthosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfc19newadmhosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfflunewadmpedhosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "totalconfflunewadmadulthosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "totalconfflunewadmhosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfrsvnewadmpedhosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfrsvnewadmadulthosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfrsvnewadmhosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "pctinptbedsocchosprep": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsocchosprep": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "numinptbedsperchosprep": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numinptbedsoccperchosprep": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numicubedsperchosprep": [random.uniform(0, 500) for _ in range(size)], + "numicubedsoccperchosprep": [ + random.uniform(0, 500) for _ in range(size) + ], + "totalconfc19hosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluhosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvhosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19icupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluicupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvicupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "numinptbedsperchosprepabschg": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numinptbedsoccperchospre": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numicubedsperchosprepabschg": [ + random.uniform(0, 500) for _ in range(size) + ], + "numicubedsoccperchosprepabschg": [ + random.uniform(0, 500) for _ in range(size) + ], + "totalconfc19hosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluhosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvhosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19icupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluicupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvicupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccperchospre": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccperchosprepabschg": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadult": [random.uniform(0, 100) for _ in range(size)], + "pctinptbedsoccped": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsoccadult": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsoccped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19icubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19hosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "totalconfc19newadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfc19newadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfc19newadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctinptbedsoccadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccpedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccpedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccpedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccpedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctinptbedsoccpedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pcticubedsoccadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pcticubedsoccpedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19inptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19inptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19icubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19icubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluinptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluinptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluicubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluicubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvicubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvicubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "respseason": [None for _ in range(size)], + "totalconfc19newadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfflunewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfrsvnewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfnewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "numconfc19icupatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped5to17": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19hosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmped0to4": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconffluhosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconffluhosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvhosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfrsvhosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfrsvhosppats": [ + random.uniform(0, 60000) for _ in range(size) + ], + "numconffluicupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconffluicupatsped": [random.uniform(0, 5000) for _ in range(size)], + "numconfrsvicupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvicupatsped": [random.uniform(0, 5000) for _ in range(size)], + "totalconfrsvicupats": [random.uniform(0, 15000) for _ in range(size)], + "numconfc19newadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmped0to4": [ + random.uniform(0, 5000) for _ in range(size) + ], + "numconfflunewadmped5to17": [ + random.uniform(0, 5000) for _ in range(size) + ], + "totalconfflunewadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult18to49": [ + random.uniform(0, 15000) for _ in range(size) + ], + "numconfflunewadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfflunewadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfflunewadmunk": [random.uniform(0, 5000) for _ in range(size)], + "numconfrsvnewadmped0to4": [ + random.uniform(0, 5000) for _ in range(size) + ], + "numconfrsvnewadmped5to17": [ + random.uniform(0, 5000) for _ in range(size) + ], + "totalconfrsvnewadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult18to49": [ + random.uniform(0, 15000) for _ in range(size) + ], + "numconfrsvnewadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfrsvnewadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfrsvnewadmunk": [random.uniform(0, 5000) for _ in range(size)], + "totalconfrsvnewadm": [random.uniform(0, 40000) for _ in range(size)], + "pctconfrsvinptbeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvicubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfflunewadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfflunewadmped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvnewadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvnewadmped": [random.uniform(0, 100) for _ in range(size)], + "numconfc19newadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconffluicubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvinptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvicubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconffluhosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvhosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + } + df = pd.DataFrame(data) + return df if output == "pandas" or output == "pd" else pl.from_pandas(df) + + +def load_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: + data = { + "weekendingdate": [str(fake.date()) for _ in range(size)], + "jurisdiction": [random.choice(region_opts) for _ in range(size)], + "numinptbeds": [random.uniform(7, 831870) for _ in range(size)], + "numinptbedsadult": [random.uniform(3, 743870) for _ in range(size)], + "numinptbedsped": [random.uniform(0, 84581) for _ in range(size)], + "numinptbedsocc": [random.uniform(0, 587461) for _ in range(size)], + "numinptbedsoccadult": [ + random.uniform(0, 543870) for _ in range(size) + ], + "numinptbedsoccped": [random.uniform(0, 84581) for _ in range(size)], + "numicubeds": [random.uniform(0, 100000) for _ in range(size)], + "numicubedsadult": [random.uniform(0, 90000) for _ in range(size)], + "numicubedsped": [random.uniform(0, 10000) for _ in range(size)], + "numicubedsocc": [random.uniform(0, 90000) for _ in range(size)], + "numicubedsoccadult": [random.uniform(0, 80000) for _ in range(size)], + "numicubedsoccped": [random.uniform(0, 10000) for _ in range(size)], + "numconfc19hosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfc19hosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19hosppats": [ + random.uniform(0, 60000) for _ in range(size) + ], + "totalconffluhosppats": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfc19icupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19icupats": [random.uniform(0, 15000) for _ in range(size)], + "totalconffluicupats": [random.uniform(0, 12000) for _ in range(size)], + "totalconfc19newadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult18to49": [ + random.uniform(0, 20000) for _ in range(size) + ], + "totalconfc19newadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfc19newadmunk": [random.uniform(0, 5000) for _ in range(size)], + "totalconfc19newadm": [random.uniform(0, 40000) for _ in range(size)], + "totalconfflunewadm": [random.uniform(0, 35000) for _ in range(size)], + "pctinptbedsocc": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbeds": [random.uniform(0, 100) for _ in range(size)], + "pctconffluinptbeds": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsocc": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19icubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconffluicubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19newadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19newadmped": [random.uniform(0, 100) for _ in range(size)], + "numinptbedshosprep": [random.uniform(0, 500000) for _ in range(size)], + "numinptbedsocchosprep": [ + random.uniform(0, 400000) for _ in range(size) + ], + "numicubedshosprep": [random.uniform(0, 100000) for _ in range(size)], + "numicubedsocchosprep": [ + random.uniform(0, 90000) for _ in range(size) + ], + "totalconfc19hosppatshosprep": [ + random.uniform(0, 60000) for _ in range(size) + ], + "totalconffluhosppatshosprep": [ + random.uniform(0, 50000) for _ in range(size) + ], + "totalconfrsvhosppatshosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfc19icupatshosprep": [ + random.uniform(0, 15000) for _ in range(size) + ], + "totalconffluicupatshosprep": [ + random.uniform(0, 12000) for _ in range(size) + ], + "totalconfrsvicupatshosprep": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19newadmpedhosprep": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19newadmadulthosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfc19newadmhosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfflunewadmpedhosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "totalconfflunewadmadulthosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "totalconfflunewadmhosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfrsvnewadmpedhosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfrsvnewadmadulthosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfrsvnewadmhosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "pctinptbedsocchosprep": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsocchosprep": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "numinptbedsperchosprep": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numinptbedsoccperchosprep": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numicubedsperchosprep": [random.uniform(0, 500) for _ in range(size)], + "numicubedsoccperchosprep": [ + random.uniform(0, 500) for _ in range(size) + ], + "totalconfc19hosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluhosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvhosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19icupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluicupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvicupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "numinptbedsperchosprepabschg": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numinptbedsoccperchospre": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numicubedsperchosprepabschg": [ + random.uniform(0, 500) for _ in range(size) + ], + "numicubedsoccperchosprepabschg": [ + random.uniform(0, 500) for _ in range(size) + ], + "totalconfc19hosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluhosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvhosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19icupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluicupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvicupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccperchospre": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccperchosprepabschg": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadult": [random.uniform(0, 100) for _ in range(size)], + "pctinptbedsoccped": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsoccadult": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsoccped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19icubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19hosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "totalconfc19newadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfc19newadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfc19newadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctinptbedsoccadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccpedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccpedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccpedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccpedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctinptbedsoccpedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pcticubedsoccadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pcticubedsoccpedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19inptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19inptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19icubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19icubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluinptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluinptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluicubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluicubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvicubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvicubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "respseason": [None for _ in range(size)], + "totalconfc19newadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfflunewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfrsvnewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfnewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "numconfc19icupatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped5to17": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19hosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmped0to4": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconffluhosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconffluhosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvhosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfrsvhosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfrsvhosppats": [ + random.uniform(0, 60000) for _ in range(size) + ], + "numconffluicupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconffluicupatsped": [random.uniform(0, 5000) for _ in range(size)], + "numconfrsvicupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvicupatsped": [random.uniform(0, 5000) for _ in range(size)], + "totalconfrsvicupats": [random.uniform(0, 15000) for _ in range(size)], + "numconfc19newadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmped0to4": [ + random.uniform(0, 5000) for _ in range(size) + ], + "numconfflunewadmped5to17": [ + random.uniform(0, 5000) for _ in range(size) + ], + "totalconfflunewadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult18to49": [ + random.uniform(0, 15000) for _ in range(size) + ], + "numconfflunewadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfflunewadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfflunewadmunk": [random.uniform(0, 5000) for _ in range(size)], + "numconfrsvnewadmped0to4": [ + random.uniform(0, 5000) for _ in range(size) + ], + "numconfrsvnewadmped5to17": [ + random.uniform(0, 5000) for _ in range(size) + ], + "totalconfrsvnewadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult18to49": [ + random.uniform(0, 15000) for _ in range(size) + ], + "numconfrsvnewadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfrsvnewadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfrsvnewadmunk": [random.uniform(0, 5000) for _ in range(size)], + "totalconfrsvnewadm": [random.uniform(0, 40000) for _ in range(size)], + "pctconfrsvinptbeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvicubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfflunewadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfflunewadmped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvnewadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvnewadmped": [random.uniform(0, 100) for _ in range(size)], + "numconfc19newadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconffluicubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvinptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvicubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconffluhosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvhosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + } + df = pd.DataFrame(data) + return df if output == "pandas" or output == "pd" else pl.from_pandas(df) diff --git a/cfa/catalog/public/datasets/stf/schemas/nhsn_hrd_prelim.py b/cfa/catalog/public/datasets/stf/schemas/nhsn_hrd_prelim.py new file mode 100644 index 0000000..efff9dd --- /dev/null +++ b/cfa/catalog/public/datasets/stf/schemas/nhsn_hrd_prelim.py @@ -0,0 +1,3436 @@ +import random + +import pandas as pd +import pandera.pandas as pa +import polars as pl +from faker import Faker + +fake = Faker() + +region_opts = [ + "AL", + "AR", + "AS", + "AZ", + "CA", + "CO", + "CT", + "DC", + "DE", + "FL", + "GU", + "HI", + "IL", + "IN", + "KS", + "ME", + "MN", + "MP", + "MS", + "MT", + "NE", + "NH", + "OH", + "OK", + "PA", + "PR", + "RI", + "Region 10", + "Region 2", + "Region 4", + "Region 6", + "Region 7", + "SC", + "SD", + "TN", + "USA", + "WI", + "ID", + "LA", + "MD", + "NJ", + "NM", + "OR", + "Region 8", + "VT", + "WA", + "AK", + "GA", + "MO", + "NV", + "Region 5", + "Region 9", + "UT", + "KY", + "MA", + "MI", + "NC", + "Region 1", + "TX", + "WY", + "Region 3", + "VA", + "VI", + "WV", + "NY", + "IA", + "ND", +] + + +extract_schema = pa.DataFrameSchema( + { + "weekendingdate": pa.Column(object, nullable=True, coerce=True), + "jurisdiction": pa.Column( + str, pa.Check.isin(region_opts), nullable=True, coerce=True + ), + "numinptbeds": pa.Column(float, nullable=True, coerce=True), + "numinptbedsadult": pa.Column(float, nullable=True, coerce=True), + "numinptbedsped": pa.Column(float, nullable=True, coerce=True), + "numinptbedsocc": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccadult": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccped": pa.Column(float, nullable=True, coerce=True), + "numicubeds": pa.Column(float, nullable=True, coerce=True), + "numicubedsadult": pa.Column(float, nullable=True, coerce=True), + "numicubedsped": pa.Column(float, nullable=True, coerce=True), + "numicubedsocc": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccadult": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccped": pa.Column(float, nullable=True, coerce=True), + "numconfc19hosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19hosppatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfc19hosppats": pa.Column(float, nullable=True, coerce=True), + "totalconffluhosppats": pa.Column(float, nullable=True, coerce=True), + "numconfc19icupatsadult": pa.Column(float, nullable=True, coerce=True), + "totalconfc19icupats": pa.Column(float, nullable=True, coerce=True), + "totalconffluicupats": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadmped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmunk": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadm": pa.Column(float, nullable=True, coerce=True), + "totalconfflunewadm": pa.Column(float, nullable=True, coerce=True), + "pctinptbedsocc": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbeds": pa.Column(float, nullable=True, coerce=True), + "pctconffluinptbeds": pa.Column(float, nullable=True, coerce=True), + "pcticubedsocc": pa.Column(float, nullable=True, coerce=True), + "pctconfc19icubeds": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubeds": pa.Column(float, nullable=True, coerce=True), + "pctconfc19newadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfc19newadmped": pa.Column(float, nullable=True, coerce=True), + "numinptbedshosprep": pa.Column(float, nullable=True, coerce=True), + "numinptbedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedshosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "totalconfc19hosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "pcticubedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsperchosprep": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsperchosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19hosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsoccperchospre": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsoccperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19hosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccperchospre": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadult": pa.Column(float, nullable=True, coerce=True), + "pctinptbedsoccped": pa.Column(float, nullable=True, coerce=True), + "pcticubedsoccadult": pa.Column(float, nullable=True, coerce=True), + "pcticubedsoccped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19icubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicu": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicu": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "respseason": pa.Column(str, nullable=True, coerce=True), + "totalconfc19newadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfnewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19icupatsped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconffluhosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconffluhosppatsped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvhosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvhosppatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvhosppats": pa.Column(float, nullable=True, coerce=True), + "numconffluicupatsadult": pa.Column(float, nullable=True, coerce=True), + "numconffluicupatsped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvicupatsadult": pa.Column(float, nullable=True, coerce=True), + "numconfrsvicupatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvicupats": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfflunewadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmunk": pa.Column(float, nullable=True, coerce=True), + "numconfrsvnewadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvnewadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmunk": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvnewadm": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvinptbeds": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubeds": pa.Column(float, nullable=True, coerce=True), + "pctconfflunewadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfflunewadmped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvnewadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvnewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvinptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconffluhosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicu": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvhosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + } +) + + +load_schema = pa.DataFrameSchema( + { + "weekendingdate": pa.Column(str, nullable=True, coerce=True), + "jurisdiction": pa.Column( + str, pa.Check.isin(region_opts), nullable=True, coerce=True + ), + "numinptbeds": pa.Column(float, nullable=True, coerce=True), + "numinptbedsadult": pa.Column(float, nullable=True, coerce=True), + "numinptbedsped": pa.Column(float, nullable=True, coerce=True), + "numinptbedsocc": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccadult": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccped": pa.Column(float, nullable=True, coerce=True), + "numicubeds": pa.Column(float, nullable=True, coerce=True), + "numicubedsadult": pa.Column(float, nullable=True, coerce=True), + "numicubedsped": pa.Column(float, nullable=True, coerce=True), + "numicubedsocc": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccadult": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccped": pa.Column(float, nullable=True, coerce=True), + "numconfc19hosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19hosppatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfc19hosppats": pa.Column(float, nullable=True, coerce=True), + "totalconffluhosppats": pa.Column(float, nullable=True, coerce=True), + "numconfc19icupatsadult": pa.Column(float, nullable=True, coerce=True), + "totalconfc19icupats": pa.Column(float, nullable=True, coerce=True), + "totalconffluicupats": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadmped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmunk": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadm": pa.Column(float, nullable=True, coerce=True), + "totalconfflunewadm": pa.Column(float, nullable=True, coerce=True), + "pctinptbedsocc": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbeds": pa.Column(float, nullable=True, coerce=True), + "pctconffluinptbeds": pa.Column(float, nullable=True, coerce=True), + "pcticubedsocc": pa.Column(float, nullable=True, coerce=True), + "pctconfc19icubeds": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubeds": pa.Column(float, nullable=True, coerce=True), + "pctconfc19newadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfc19newadmped": pa.Column(float, nullable=True, coerce=True), + "numinptbedshosprep": pa.Column(float, nullable=True, coerce=True), + "numinptbedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedshosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "totalconfc19hosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "pcticubedsocchosprep": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedshosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsperchosprep": pa.Column(float, nullable=True, coerce=True), + "numinptbedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsperchosprep": pa.Column(float, nullable=True, coerce=True), + "numicubedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19hosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatsperc": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultp": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "numinptbedsoccperchospre": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "numicubedsoccperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19hosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluhosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvhosppatsperc_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19icupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconffluicupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvicupatsperch": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultp_1": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccperchospre": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspercho": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsperchos": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprepabove80pct": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmperchosprepabove90pct": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadult": pa.Column(float, nullable=True, coerce=True), + "pctinptbedsoccped": pa.Column(float, nullable=True, coerce=True), + "pcticubedsoccadult": pa.Column(float, nullable=True, coerce=True), + "pcticubedsoccped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19inptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19icubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicu": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicu": pa.Column(float, nullable=True, coerce=True), + "totalconfc19newadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfc19newadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadulthosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedhosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadultperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedperchosprep": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctinptbedsoccpedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pcticubedsoccpedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19inptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluicubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedsadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvicubedspedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19hosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuadultperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicupedperchosprepabschg": pa.Column( + float, nullable=True, coerce=True + ), + "respseason": pa.Column(str, nullable=True, coerce=True), + "totalconfc19newadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfnewadmcumulativeseasonalsum": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19icupatsped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfc19icubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfc19hosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconffluhosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconffluhosppatsped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvhosppatsadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvhosppatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvhosppats": pa.Column(float, nullable=True, coerce=True), + "numconffluicupatsadult": pa.Column(float, nullable=True, coerce=True), + "numconffluicupatsped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvicupatsadult": pa.Column(float, nullable=True, coerce=True), + "numconfrsvicupatsped": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvicupats": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfflunewadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmunk": pa.Column(float, nullable=True, coerce=True), + "numconfrsvnewadmped0to4": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfrsvnewadmadult18to49": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75plus": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadult": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmunk": pa.Column(float, nullable=True, coerce=True), + "totalconfrsvnewadm": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvinptbeds": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubeds": pa.Column(float, nullable=True, coerce=True), + "pctconfflunewadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfflunewadmped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvnewadmadult": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvnewadmped": pa.Column(float, nullable=True, coerce=True), + "numconfc19newadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped0to4per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17per100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedper100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult18to49per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74per100k": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75plusper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultper100k": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmper100k": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluinptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconffluicubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvinptbedsadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvinptbedsped": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubedsadult": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvicubedsped": pa.Column(float, nullable=True, coerce=True), + "pctconffluhosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicu": pa.Column(float, nullable=True, coerce=True), + "pctconfrsvhosppatsicuadult": pa.Column( + float, nullable=True, coerce=True + ), + "pctconfrsvhosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfc19newadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfflunewadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmped5to17pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpedpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult18to49pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult50to64pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmadultpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "totalconfrsvnewadmpctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult75pluspctchg": pa.Column( + float, nullable=True, coerce=True + ), + "numconfrsvnewadmadult65to74pctchg": pa.Column( + float, nullable=True, coerce=True + ), + "pctconffluhosppatsicuped": pa.Column( + float, nullable=True, coerce=True + ), + "numconfflunewadmped0to4pctchg": pa.Column( + float, nullable=True, coerce=True + ), + } +) + + +def extract_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: + data = { + "weekendingdate": [fake.date() for _ in range(size)], + "jurisdiction": [random.choice(region_opts) for _ in range(size)], + "numinptbeds": [random.uniform(7, 831870) for _ in range(size)], + "numinptbedsadult": [random.uniform(3, 743870) for _ in range(size)], + "numinptbedsped": [random.uniform(0, 84581) for _ in range(size)], + "numinptbedsocc": [random.uniform(0, 587461) for _ in range(size)], + "numinptbedsoccadult": [ + random.uniform(0, 543870) for _ in range(size) + ], + "numinptbedsoccped": [random.uniform(0, 84581) for _ in range(size)], + "numicubeds": [random.uniform(0, 100000) for _ in range(size)], + "numicubedsadult": [random.uniform(0, 90000) for _ in range(size)], + "numicubedsped": [random.uniform(0, 10000) for _ in range(size)], + "numicubedsocc": [random.uniform(0, 90000) for _ in range(size)], + "numicubedsoccadult": [random.uniform(0, 80000) for _ in range(size)], + "numicubedsoccped": [random.uniform(0, 10000) for _ in range(size)], + "numconfc19hosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfc19hosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19hosppats": [ + random.uniform(0, 60000) for _ in range(size) + ], + "totalconffluhosppats": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfc19icupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19icupats": [random.uniform(0, 15000) for _ in range(size)], + "totalconffluicupats": [random.uniform(0, 12000) for _ in range(size)], + "totalconfc19newadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult18to49": [ + random.uniform(0, 20000) for _ in range(size) + ], + "totalconfc19newadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfc19newadmunk": [random.uniform(0, 5000) for _ in range(size)], + "totalconfc19newadm": [random.uniform(0, 40000) for _ in range(size)], + "totalconfflunewadm": [random.uniform(0, 35000) for _ in range(size)], + "pctinptbedsocc": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbeds": [random.uniform(0, 100) for _ in range(size)], + "pctconffluinptbeds": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsocc": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19icubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconffluicubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19newadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19newadmped": [random.uniform(0, 100) for _ in range(size)], + "numinptbedshosprep": [random.uniform(0, 500000) for _ in range(size)], + "numinptbedsocchosprep": [ + random.uniform(0, 400000) for _ in range(size) + ], + "numicubedshosprep": [random.uniform(0, 100000) for _ in range(size)], + "numicubedsocchosprep": [ + random.uniform(0, 90000) for _ in range(size) + ], + "totalconfc19hosppatshosprep": [ + random.uniform(0, 60000) for _ in range(size) + ], + "totalconffluhosppatshosprep": [ + random.uniform(0, 50000) for _ in range(size) + ], + "totalconfrsvhosppatshosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfc19icupatshosprep": [ + random.uniform(0, 15000) for _ in range(size) + ], + "totalconffluicupatshosprep": [ + random.uniform(0, 12000) for _ in range(size) + ], + "totalconfrsvicupatshosprep": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19newadmpedhosprep": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19newadmadulthosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfc19newadmhosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfflunewadmpedhosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "totalconfflunewadmadulthosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "totalconfflunewadmhosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfrsvnewadmpedhosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfrsvnewadmadulthosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfrsvnewadmhosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "pctinptbedsocchosprep": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsocchosprep": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "numinptbedsperchosprep": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numinptbedsoccperchosprep": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numicubedsperchosprep": [random.uniform(0, 500) for _ in range(size)], + "numicubedsoccperchosprep": [ + random.uniform(0, 500) for _ in range(size) + ], + "totalconfc19hosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluhosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvhosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19icupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluicupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvicupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "numinptbedsperchosprepabschg": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numinptbedsoccperchospre": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numicubedsperchosprepabschg": [ + random.uniform(0, 500) for _ in range(size) + ], + "numicubedsoccperchosprepabschg": [ + random.uniform(0, 500) for _ in range(size) + ], + "totalconfc19hosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluhosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvhosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19icupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluicupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvicupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccperchospre": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccperchosprepabschg": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadult": [random.uniform(0, 100) for _ in range(size)], + "pctinptbedsoccped": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsoccadult": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsoccped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19icubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19hosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "totalconfc19newadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfc19newadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfc19newadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctinptbedsoccadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccpedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccpedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccpedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccpedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctinptbedsoccpedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pcticubedsoccadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pcticubedsoccpedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19inptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19inptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19icubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19icubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluinptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluinptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluicubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluicubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvicubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvicubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "respseason": [None for _ in range(size)], + "totalconfc19newadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfflunewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfrsvnewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfnewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "numconfc19icupatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped5to17": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19hosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmped0to4": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconffluhosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconffluhosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvhosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfrsvhosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfrsvhosppats": [ + random.uniform(0, 60000) for _ in range(size) + ], + "numconffluicupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconffluicupatsped": [random.uniform(0, 5000) for _ in range(size)], + "numconfrsvicupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvicupatsped": [random.uniform(0, 5000) for _ in range(size)], + "totalconfrsvicupats": [random.uniform(0, 15000) for _ in range(size)], + "numconfc19newadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmped0to4": [ + random.uniform(0, 5000) for _ in range(size) + ], + "numconfflunewadmped5to17": [ + random.uniform(0, 5000) for _ in range(size) + ], + "totalconfflunewadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult18to49": [ + random.uniform(0, 15000) for _ in range(size) + ], + "numconfflunewadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfflunewadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfflunewadmunk": [random.uniform(0, 5000) for _ in range(size)], + "numconfrsvnewadmped0to4": [ + random.uniform(0, 5000) for _ in range(size) + ], + "numconfrsvnewadmped5to17": [ + random.uniform(0, 5000) for _ in range(size) + ], + "totalconfrsvnewadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult18to49": [ + random.uniform(0, 15000) for _ in range(size) + ], + "numconfrsvnewadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfrsvnewadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfrsvnewadmunk": [random.uniform(0, 5000) for _ in range(size)], + "totalconfrsvnewadm": [random.uniform(0, 40000) for _ in range(size)], + "pctconfrsvinptbeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvicubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfflunewadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfflunewadmped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvnewadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvnewadmped": [random.uniform(0, 100) for _ in range(size)], + "numconfc19newadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconffluicubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvinptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvicubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconffluhosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvhosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + } + df = pd.DataFrame(data) + return df if output == "pandas" or output == "pd" else pl.from_pandas(df) + + +def load_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: + data = { + "weekendingdate": [str(fake.date()) for _ in range(size)], + "jurisdiction": [random.choice(region_opts) for _ in range(size)], + "numinptbeds": [random.uniform(7, 831870) for _ in range(size)], + "numinptbedsadult": [random.uniform(3, 743870) for _ in range(size)], + "numinptbedsped": [random.uniform(0, 84581) for _ in range(size)], + "numinptbedsocc": [random.uniform(0, 587461) for _ in range(size)], + "numinptbedsoccadult": [ + random.uniform(0, 543870) for _ in range(size) + ], + "numinptbedsoccped": [random.uniform(0, 84581) for _ in range(size)], + "numicubeds": [random.uniform(0, 100000) for _ in range(size)], + "numicubedsadult": [random.uniform(0, 90000) for _ in range(size)], + "numicubedsped": [random.uniform(0, 10000) for _ in range(size)], + "numicubedsocc": [random.uniform(0, 90000) for _ in range(size)], + "numicubedsoccadult": [random.uniform(0, 80000) for _ in range(size)], + "numicubedsoccped": [random.uniform(0, 10000) for _ in range(size)], + "numconfc19hosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfc19hosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19hosppats": [ + random.uniform(0, 60000) for _ in range(size) + ], + "totalconffluhosppats": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfc19icupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19icupats": [random.uniform(0, 15000) for _ in range(size)], + "totalconffluicupats": [random.uniform(0, 12000) for _ in range(size)], + "totalconfc19newadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult18to49": [ + random.uniform(0, 20000) for _ in range(size) + ], + "totalconfc19newadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfc19newadmunk": [random.uniform(0, 5000) for _ in range(size)], + "totalconfc19newadm": [random.uniform(0, 40000) for _ in range(size)], + "totalconfflunewadm": [random.uniform(0, 35000) for _ in range(size)], + "pctinptbedsocc": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbeds": [random.uniform(0, 100) for _ in range(size)], + "pctconffluinptbeds": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsocc": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19icubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconffluicubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19newadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19newadmped": [random.uniform(0, 100) for _ in range(size)], + "numinptbedshosprep": [random.uniform(0, 500000) for _ in range(size)], + "numinptbedsocchosprep": [ + random.uniform(0, 400000) for _ in range(size) + ], + "numicubedshosprep": [random.uniform(0, 100000) for _ in range(size)], + "numicubedsocchosprep": [ + random.uniform(0, 90000) for _ in range(size) + ], + "totalconfc19hosppatshosprep": [ + random.uniform(0, 60000) for _ in range(size) + ], + "totalconffluhosppatshosprep": [ + random.uniform(0, 50000) for _ in range(size) + ], + "totalconfrsvhosppatshosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfc19icupatshosprep": [ + random.uniform(0, 15000) for _ in range(size) + ], + "totalconffluicupatshosprep": [ + random.uniform(0, 12000) for _ in range(size) + ], + "totalconfrsvicupatshosprep": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19newadmpedhosprep": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfc19newadmadulthosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfc19newadmhosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfflunewadmpedhosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "totalconfflunewadmadulthosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "totalconfflunewadmhosprep": [ + random.uniform(0, 40000) for _ in range(size) + ], + "totalconfrsvnewadmpedhosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfrsvnewadmadulthosprep": [ + random.uniform(0, 30000) for _ in range(size) + ], + "totalconfrsvnewadmhosprep": [ + random.uniform(0, 35000) for _ in range(size) + ], + "pctinptbedsocchosprep": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsocchosprep": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedshosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "numinptbedsperchosprep": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numinptbedsoccperchosprep": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numicubedsperchosprep": [random.uniform(0, 500) for _ in range(size)], + "numicubedsoccperchosprep": [ + random.uniform(0, 500) for _ in range(size) + ], + "totalconfc19hosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluhosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvhosppatsperc": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19icupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluicupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvicupatsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultp": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "numinptbedsperchosprepabschg": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numinptbedsoccperchospre": [ + random.uniform(0, 1000) for _ in range(size) + ], + "numicubedsperchosprepabschg": [ + random.uniform(0, 500) for _ in range(size) + ], + "numicubedsoccperchosprepabschg": [ + random.uniform(0, 500) for _ in range(size) + ], + "totalconfc19hosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluhosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvhosppatsperc_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19icupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconffluicupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvicupatsperch": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultp_1": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccperchospre": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccperchosprepabschg": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspercho": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsperchos": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfc19newadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprepabove80pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmperchosprepabove90pct": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadult": [random.uniform(0, 100) for _ in range(size)], + "pctinptbedsoccped": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsoccadult": [random.uniform(0, 100) for _ in range(size)], + "pcticubedsoccped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19inptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19icubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19hosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "totalconfc19newadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfc19newadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfc19newadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctinptbedsoccadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccpedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccpedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedspedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadulthosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedhosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccpedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pcticubedsoccpedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19inptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedspedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadultperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedperchosprep": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctinptbedsoccadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctinptbedsoccpedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pcticubedsoccadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pcticubedsoccpedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19inptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19inptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19icubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19icubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluinptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluinptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluicubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluicubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvinptbedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvinptbedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvicubedsadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvicubedspedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfc19hosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuadultperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicupedperchosprepabschg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "respseason": [None for _ in range(size)], + "totalconfc19newadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfflunewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfrsvnewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "totalconfnewadmcumulativeseasonalsum": [ + random.uniform(0, 100000) for _ in range(size) + ], + "numconfc19icupatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped5to17": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfc19icubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfc19hosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmped0to4": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconffluhosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconffluhosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvhosppatsadult": [ + random.uniform(0, 50000) for _ in range(size) + ], + "numconfrsvhosppatsped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfrsvhosppats": [ + random.uniform(0, 60000) for _ in range(size) + ], + "numconffluicupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconffluicupatsped": [random.uniform(0, 5000) for _ in range(size)], + "numconfrsvicupatsadult": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvicupatsped": [random.uniform(0, 5000) for _ in range(size)], + "totalconfrsvicupats": [random.uniform(0, 15000) for _ in range(size)], + "numconfc19newadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfc19newadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmped0to4": [ + random.uniform(0, 5000) for _ in range(size) + ], + "numconfflunewadmped5to17": [ + random.uniform(0, 5000) for _ in range(size) + ], + "totalconfflunewadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult18to49": [ + random.uniform(0, 15000) for _ in range(size) + ], + "numconfflunewadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfflunewadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfflunewadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfflunewadmunk": [random.uniform(0, 5000) for _ in range(size)], + "numconfrsvnewadmped0to4": [ + random.uniform(0, 5000) for _ in range(size) + ], + "numconfrsvnewadmped5to17": [ + random.uniform(0, 5000) for _ in range(size) + ], + "totalconfrsvnewadmped": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult18to49": [ + random.uniform(0, 15000) for _ in range(size) + ], + "numconfrsvnewadmadult50to64": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult65to74": [ + random.uniform(0, 10000) for _ in range(size) + ], + "numconfrsvnewadmadult75plus": [ + random.uniform(0, 10000) for _ in range(size) + ], + "totalconfrsvnewadmadult": [ + random.uniform(0, 30000) for _ in range(size) + ], + "numconfrsvnewadmunk": [random.uniform(0, 5000) for _ in range(size)], + "totalconfrsvnewadm": [random.uniform(0, 40000) for _ in range(size)], + "pctconfrsvinptbeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvicubeds": [random.uniform(0, 100) for _ in range(size)], + "pctconfflunewadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfflunewadmped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvnewadmadult": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvnewadmped": [random.uniform(0, 100) for _ in range(size)], + "numconfc19newadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfflunewadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmped0to4per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmped5to17per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult18to49per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult50to64per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult65to74per100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfrsvnewadmadult75plusper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "totalconfrsvnewadmper100k": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluinptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconffluicubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconffluicubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvinptbedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvinptbedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvicubedsadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvicubedsped": [random.uniform(0, 100) for _ in range(size)], + "pctconffluhosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicu": [random.uniform(0, 100) for _ in range(size)], + "pctconfrsvhosppatsicuadult": [ + random.uniform(0, 100) for _ in range(size) + ], + "pctconfrsvhosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfc19newadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfc19newadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfflunewadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfflunewadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmped5to17pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmpedpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult18to49pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult50to64pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmadultpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "totalconfrsvnewadmpctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult75pluspctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "numconfrsvnewadmadult65to74pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + "pctconffluhosppatsicuped": [ + random.uniform(0, 100) for _ in range(size) + ], + "numconfflunewadmped0to4pctchg": [ + random.uniform(-100, 100) for _ in range(size) + ], + } + df = pd.DataFrame(data) + return df if output == "pandas" or output == "pd" else pl.from_pandas(df) diff --git a/cfa/catalog/public/datasets/stf/schemas/nssp_gold_v1.py b/cfa/catalog/public/datasets/stf/schemas/nssp_gold_v1.py new file mode 100644 index 0000000..d4264e0 --- /dev/null +++ b/cfa/catalog/public/datasets/stf/schemas/nssp_gold_v1.py @@ -0,0 +1,122 @@ +import random + +import pandas as pd +import pandera.pandas as pa +import polars as pl + +metric_opts = [ + "count_ed_visits", + "percent_ed_visits", + "count_admitted_ed_visits", +] +geo_value_opts = [ + "AK", + "AL", + "AR", + "AZ", + "CA", + "CO", + "CT", + "DC", + "DE", + "FL", + "GA", + "GU", + "HI", + "IA", + "ID", + "IL", + "IN", + "KS", + "KY", + "LA", + "MA", + "MD", + "ME", + "MI", + "MN", + "MO", + "MS", + "MT", + "NC", + "ND", + "NE", + "NH", + "NJ", + "NM", + "NV", + "NY", + "OH", + "OK", + "OR", + "PA", + "RI", + "SC", + "SD", + "TN", + "TX", + "UT", + "VA", + "VT", + "WA", + "WI", + "WV", + "WY", +] +disease_opts = ["COVID-19/Omicron", "RSV", "Influenza", "Total"] + +load_schema = pa.DataFrameSchema( + { + "report_date": pa.Column( + str, + pa.Check.str_matches( + r"^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}(\.\d{3})?$" + ), + coerce=True, + ), # Check if report_date can be converted to datetime + "reference_date": pa.Column( + str, pa.Check(lambda x: pd.to_datetime(x, errors="coerce").notna()) + ), # Check if reference_date can be converted to datetime + "asof": pa.Column( + str, pa.Check(lambda x: pd.to_datetime(x, errors="coerce").notna()) + ), # Check if asof can be converted to datetime + "facility": pa.Column( + int, pa.Check(lambda x: x >= 0) + ), # Check if facility is a non-negative integer + "metric": pa.Column( + str, pa.Check.isin(metric_opts) + ), # Check if metric is one of the predefined options + "geo_type": pa.Column( + str, pa.Check.isin(["state"]) + ), # Check if geo_type is a non-empty string + "geo_value": pa.Column( + str, pa.Check.isin(geo_value_opts) + ), # Check if geo_value is one of the predefined options + "run_id": pa.Column( + str + ), # Check if run_id is one of the predefined options + "disease": pa.Column( + str, pa.Check.isin(disease_opts) + ), # Check if disease is one of the predefined options + "value": pa.Column( + float, pa.Check(lambda x: x >= 0) + ), # Check if value is a non-negative float + } +) + + +def load_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: + df = pd.DataFrame() + df = df.assign( + report_date=["2026-02-02T00:00:00.000"] * size, + reference_date=["2023-07-01T00:00:00.000"] * size, + asof=["2026-02-02T00:00:00.000"] * size, + facility=[random.randint(1237, 35720) for _ in range(size)], + metric=random.choices(metric_opts, k=size), + geo_type=["state"] * size, + geo_value=random.choices(geo_value_opts, k=size), + run_id=["598084b2-5f3a-4d07-b473-b7903ed7e669" for _ in range(size)], + disease=random.choices(disease_opts, k=size), + value=[random.uniform(0, 30000) for _ in range(size)], + ) + return df if output == "pandas" or output == "pd" else pl.from_pandas(df) diff --git a/cfa/catalog/public/datasets/stf/schemas/nssp_latest_comprehensive.py b/cfa/catalog/public/datasets/stf/schemas/nssp_latest_comprehensive.py deleted file mode 100644 index d08ba31..0000000 --- a/cfa/catalog/public/datasets/stf/schemas/nssp_latest_comprehensive.py +++ /dev/null @@ -1,111 +0,0 @@ -import random - -import pandas as pd -import pandera.pandas as pa -import polars as pl -from faker import Faker - -fake = Faker() - -metric_opts = [ - "count_ed_visits", - "count_admitted_ed_visits", - "percent_ed_visits", -] -state_opts = [ - "FL", - "RI", - "MO", - "NJ", - "CO", - "NM", - "VT", - "MA", - "MD", - "WY", - "UT", - "ME", - "SC", - "WI", - "TN", - "NY", - "MN", - "IL", - "IN", - "LA", - "MI", - "TX", - "DC", - "CT", - "ID", - "GA", - "SD", - "AK", - "WA", - "WV", - "VA", - "NE", - "IA", - "PA", - "NV", - "OK", - "AR", - "OH", - "OR", - "DE", - "KY", - "MS", - "ND", - "NC", - "NH", - "KS", - "GU", - "HI", - "AZ", - "CA", - "MT", - "AL", -] -disease_opts = ["COVID-19/Omicron", "Total", "Influenza", "RSV"] - -load_schema = pa.DataFrameSchema( - { - "report_date": pa.Column( - str, - pa.Check.str_matches(r"^\d{4}-\d{2}-\d{2}$"), - coerce=True, - ), - "reference_date": pa.Column( - str, - pa.Check.str_matches(r"^\d{4}-\d{2}-\d{2}$"), - coerce=True, - ), - "metric": pa.Column(str, pa.Check.isin(metric_opts), nullable=True), - "geo_type": pa.Column(str, pa.Check.isin(["state"]), nullable=True), - "geo_value": pa.Column(str, pa.Check.isin(state_opts), nullable=True), - "disease": pa.Column(str, pa.Check.isin(disease_opts), nullable=True), - "value": pa.Column( - float, pa.Check.greater_than_or_equal_to(0.0), nullable=True - ), - } -) - - -def load_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: - df = pd.DataFrame() - df = df.assign( - report_date=[ - fake.date_between(start_date="-1y", end_date="today").isoformat() - for _ in range(size) - ], - reference_date=[ - fake.date_between(start_date="-3y", end_date="today").isoformat() - for _ in range(size) - ], - metric=[random.choice(metric_opts) for _ in range(size)], - geo_type=["state" for _ in range(size)], - geo_value=[random.choice(state_opts) for _ in range(size)], - disease=[random.choice(disease_opts) for _ in range(size)], - value=[round(random.uniform(0, 25000), 2) for _ in range(size)], - ) - return df if output == "pandas" or output == "pd" else pl.from_pandas(df) diff --git a/cfa/catalog/public/datasets/stf/schemas/nwss.py b/cfa/catalog/public/datasets/stf/schemas/nwss.py new file mode 100644 index 0000000..65c4a56 --- /dev/null +++ b/cfa/catalog/public/datasets/stf/schemas/nwss.py @@ -0,0 +1,622 @@ +import random + +import pandas as pd +import pandera.pandas as pa +import polars as pl +from faker import Faker + +fake = Faker() + +wwtop_opts = [ + "ak", + "al", + "az", + "ca", + "ar", + "co", + "ct", + "dc", + "fl", + "ga", + "gu", + "hi", + "ia", + "id", + "il", + "in", + "ks", + "ky", + "la", + "ma", + "md", + "me", + "mn", + "mo", + "ms", + "mt", + "nc", + "nd", + "ne", + "nh", + "nj", + "nm", + "nv", + "ny", + "oh", + "ok", + "or", + "pa", + "ri", + "sc", + "sd", + "tn", + "tx", + "ut", + "va", + "vt", + "wa", + "wi", + "wy", + "de", + "mi", + "wv", +] +reporting_jur_opts = [ + "ak", + "al", + "az", + "ca", + "ar", + "lc", + "co", + "ct", + "dc", + "fl", + "ga", + "gu", + "hi", + "ia", + "id", + "il", + "ci", + "in", + "ks", + "ky", + "la", + "ma", + "md", + "me", + "mn", + "mo", + "ms", + "mt", + "nc", + "nd", + "ne", + "nh", + "nj", + "nm", + "nv", + "nz", + "ny", + "oh", + "ok", + "or", + "pa", + "ri", + "sc", + "sd", + "tn", + "tx", + "ut", + "va", + "vt", + "wa", + "wi", + "wy", + "de", + "mi", + "ph", + "ho", + "wv", +] +matrix_opts = [ + "post grit removal", + "raw wastewater", + "primary sludge", + "primary effluent", + "holding tank", + "secondary effluent", +] +sample_location_opts = ["wwtp", "upstream"] +trend_flowpop_opts = [ + "plateau", + "increase", + "decrease", + "sustained increase", + "sustained decrease", + "no clear trend", +] + +load_schema = pa.DataFrameSchema( + { + "sample_id": pa.Column(str), + "lab_id": pa.Column(str, nullable=True), + "key_plot": pa.Column(str), + "day_num": pa.Column(int, coerce=True), + "cases_new_cens_per100k": pa.Column(float, nullable=True), + "pcr_target_flowpop_lin": pa.Column(float, nullable=True), + "sample_collect_date": pa.Column( + str, + pa.Check(lambda x: pd.to_datetime(x, errors="coerce").notna()), + coerce=True, + ), + "pcr_target_flowpop_log10": pa.Column(float, nullable=True), + "rec_eff_percent": pa.Column(float, nullable=True), + "wwtp_id": pa.Column(int, nullable=True, coerce=True), + "wwtp_jurisdiction": pa.Column( + str, pa.Check.isin(wwtop_opts), nullable=True + ), + "county_names": pa.Column(str, nullable=True), + "stormwater_input": pa.Column( + str, pa.Check.isin(["yes", "no"]), nullable=True + ), + "influent_equilibrated": pa.Column( + str, pa.Check.isin(["yes", "no", "0"]), nullable=True + ), + "sample_type": pa.Column(str, nullable=True), + "sample_matrix": pa.Column( + str, pa.Check.isin(matrix_opts), nullable=True + ), + "sample_location": pa.Column( + str, pa.Check.isin(sample_location_opts), nullable=True + ), + "sample_location_specify": pa.Column( + object, nullable=True, coerce=True + ), + "reporting_jurisdiction": pa.Column( + str, pa.Check.isin(reporting_jur_opts), nullable=True + ), + "capacity_mgd": pa.Column(float, nullable=True), + "population_served": pa.Column( + str, pa.Check.isin(["yes", "no"]), nullable=True, coerce=True + ), + "industrial_input": pa.Column(float, nullable=True), + "sewage_travel_time": pa.Column(float, nullable=True), + "slopes_flowpop_5m": pa.Column(float, nullable=True), + "pvals_flowpop_5m": pa.Column(float, nullable=True), + "pdc_flowpop_5m": pa.Column(float, nullable=True), + "predsuprexc_flowpop_5m": pa.Column(float, nullable=True), + "alert_flowpop_5m": pa.Column(str, nullable=True), + "slopes_flowpop_3m": pa.Column(float, nullable=True), + "pvals_flowpop_3m": pa.Column(float, nullable=True), + "pdc_flowpop_3m": pa.Column(float, nullable=True), + "predsuprexc_flowpop_3m": pa.Column(float, nullable=True), + "alert_flowpop_3m": pa.Column(str, nullable=True), + "trend_flowpop_5m3m_p05": pa.Column(str, nullable=True), + "trend_flowpop_5m3m_p05_prev": pa.Column(str, nullable=True), + "slopes_mic_5m": pa.Column(float, nullable=True), + "pvals_mic_5m": pa.Column(float, nullable=True), + "pdc_mic_5m": pa.Column(float, nullable=True), + "predsuprexc_mic_5m": pa.Column(float, nullable=True), + "alert_mic_5m": pa.Column(str, nullable=True), + "alert_mic_3m": pa.Column(str, nullable=True), + "trend_mic_5m3m_p05": pa.Column(str, nullable=True), + "trend_mic_5m3m_p05_prev": pa.Column(str, nullable=True), + "slopes_flowpop_15d": pa.Column(float, nullable=True), + "pvals_flowpop_15d": pa.Column(float, nullable=True), + "pdc_flowpop_15d": pa.Column(float, nullable=True), + "predsuprexc_flowpop_15d": pa.Column(float, nullable=True), + "alert_flowpop_15d": pa.Column(str, nullable=True), + "slopes_flowpop_8d": pa.Column(float, nullable=True), + "pvals_flowpop_8d": pa.Column(float, nullable=True), + "pdc_flowpop_8d": pa.Column(float, nullable=True), + "predsuprexc_flowpop_8d": pa.Column(float, nullable=True), + "alert_flowpop_8d": pa.Column(str, nullable=True), + "trend_flowpop_15d8d_p05": pa.Column(str, nullable=True), + "trend_flowpop_15d8d_p05_prev": pa.Column(str, nullable=True), + "slopes_mic_15d": pa.Column(float, nullable=True), + "pvals_mic_15d": pa.Column(float, nullable=True), + "pdc_mic_15d": pa.Column(float, nullable=True), + "predsuprexc_mic_15d": pa.Column(float, nullable=True), + "alert_mic_15d": pa.Column(str, nullable=True), + "slopes_mic_8d": pa.Column(float, nullable=True), + "pvals_mic_8d": pa.Column(float, nullable=True), + "pdc_mic_8d": pa.Column(float, nullable=True), + "predsuprexc_mic_8d": pa.Column(float, nullable=True), + "alert_mic_8d": pa.Column(str, nullable=True), + "trend_mic_15d8d_p05": pa.Column(str, nullable=True), + "trend_mic_15d8d_p05_prev": pa.Column(str, nullable=True), + "cases_new_cens": pa.Column(float, nullable=True), + "pcr_target_mic_lin_roll7": pa.Column(float, nullable=True), + "pcr_target_flowpop_lin_roll7": pa.Column(float, nullable=True), + "pcr_target_flowpop_lin_spline": pa.Column(float, nullable=True), + "pcr_target_mic_lin_spline": pa.Column(float, nullable=True), + "cases_new_cens_per100k_spline": pa.Column(float, nullable=True), + "pcr_target_detect": pa.Column(str, nullable=True), + "hum_frac_mic_conc": pa.Column(float, nullable=True), + "hum_frac_mic_unit": pa.Column(str, nullable=True), + "hum_frac_chem_conc": pa.Column(float, nullable=True), + "hum_frac_chem_unit": pa.Column(str, nullable=True), + "other_norm_conc": pa.Column(float, nullable=True), + "other_norm_unit": pa.Column(str, nullable=True), + "lod_sewage": pa.Column(float, nullable=True), + "pcr_target": pa.Column(str, nullable=True), + "pcr_gene_target_agg": pa.Column(str, nullable=True), + "inhibition_detect": pa.Column(str, nullable=True), + "inhibition_adjust": pa.Column(str, nullable=True), + "hum_frac_target_chem": pa.Column(float, nullable=True), + "other_norm_name": pa.Column(str, nullable=True), + "hum_frac_target_mic": pa.Column(str, nullable=True), + "flow_rate": pa.Column(float, nullable=True), + "pcr_target_avg_conc": pa.Column(float, nullable=True), + "pcr_target_below_lod": pa.Column(str, nullable=True), + "solids_separation": pa.Column(str, nullable=True), + "concentration_method": pa.Column(str, nullable=True), + "extraction_method": pa.Column(str, nullable=True), + "rec_eff_target_name": pa.Column(str, nullable=True), + "rec_eff_spike_matrix": pa.Column(str, nullable=True), + "pasteurized": pa.Column(str, nullable=True), + "pcr_type": pa.Column(str, nullable=True), + "quant_stan_type": pa.Column(str, nullable=True), + "pcr_target_units": pa.Column(str, nullable=True), + "quality_flag": pa.Column(str, nullable=True), + "collection_storage_temp": pa.Column(float, nullable=True), + "collection_storage_time": pa.Column(float, nullable=True), + "pre_conc_storage_temp": pa.Column(float, nullable=True), + "pre_conc_storage_time": pa.Column(float, nullable=True), + "pre_ext_storage_temp": pa.Column(float, nullable=True), + "pre_ext_storage_time": pa.Column(float, nullable=True), + "sample_collect_time": pa.Column(float, nullable=True), + "time_zone": pa.Column(str, nullable=True), + "collection_water_temp": pa.Column(float, nullable=True), + "tot_conc_vol": pa.Column(float, nullable=True), + "ext_blank": pa.Column(float, nullable=True), + "num_no_target_control": pa.Column(float, nullable=True), + "tss": pa.Column(float, nullable=True), + "ph": pa.Column(float, nullable=True), + "composite_freq": pa.Column(float, nullable=True), + "pcr_target_avg_conc_log10": pa.Column(float, nullable=True), + "pcr_target_avg_conc_lin": pa.Column(float, nullable=True), + "pcr_target_mic_lin": pa.Column(float, nullable=True), + "pcr_target_mic_log10": pa.Column(float, nullable=True), + "rec_eff_spike_conc": pa.Column(float, nullable=True), + "conductivity": pa.Column(float, nullable=True), + "equiv_sewage_amt": pa.Column(float, nullable=True), + "hum_frac_mic_conc_lin": pa.Column(float, nullable=True), + "ntc_amplify": pa.Column(float, nullable=True), + "ptc_flowpop_3m": pa.Column(float, nullable=True), + "ptc_flowpop_5m": pa.Column(float, nullable=True), + "ptc_flowpop_8d": pa.Column(float, nullable=True), + "ptc_flowpop_15d": pa.Column(float, nullable=True), + "ptc_mic_3m": pa.Column(float, nullable=True), + "ptc_mic_5m": pa.Column(float, nullable=True), + "ptc_mic_8d": pa.Column(float, nullable=True), + "ptc_mic_15d": pa.Column(float, nullable=True), + "perc_flowpop": pa.Column(float, nullable=True), + "quin_flowpop_bin": pa.Column(float, nullable=True), + "perc_mic": pa.Column(float, nullable=True), + "quin_mic_bin": pa.Column(float, nullable=True), + "major_lab_method": pa.Column(str, nullable=True), + "site_id": pa.Column(int, nullable=True), + "jurisdiction_policy_rid": pa.Column(str, nullable=True), + } +) + +sample_type_opts = [ + "24-hr flow-weighted composite", + "grab", + "24-hr time-weighted composite", + "1-hr time-weighted composite", + "2-hr manual composite", + "24-hr manual composite", + "22-hr time-weighted composite", + "26-hr time-weighted composite", +] + + +def load_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: + data = { + "sample_id": [fake.uuid4() for _ in range(size)], + "lab_id": [fake.uuid4() for _ in range(size)], + "key_plot": [fake.word() for _ in range(size)], + "day_num": [random.randint(10, 2100) for _ in range(size)], + "cases_new_cens_per100k": [ + random.uniform(0, 11200) for _ in range(size) + ], + "pcr_target_flowpop_lin": [ + random.uniform(5, 4e18) for _ in range(size) + ], + "sample_collect_date": [fake.date() for _ in range(size)], + "pcr_target_flowpop_log10": [ + random.uniform(1.1, 18) for _ in range(size) + ], + "rec_eff_percent": [random.uniform(0, 100) for _ in range(size)], + "wwtp_id": [random.randint(1, 1000) for _ in range(size)], + "wwtp_jurisdiction": [random.choice(wwtop_opts) for _ in range(size)], + "county_names": [ + f"{random.randint(0, 99999):05d}" for _ in range(size) + ], + "stormwater_input": [ + random.choice(["yes", "no"]) for _ in range(size) + ], + "influent_equilibrated": [ + random.choice(["yes", "no"]) for _ in range(size) + ], + "sample_type": [random.choice(sample_type_opts) for _ in range(size)], + "sample_matrix": [random.choice(matrix_opts) for _ in range(size)], + "sample_location": [ + random.choice(sample_location_opts) for _ in range(size) + ], + "sample_location_specify": [ + random.uniform(2, 724) for _ in range(size) + ], + "reporting_jurisdiction": [ + random.choice(reporting_jur_opts) for _ in range(size) + ], + "capacity_mgd": [random.uniform(0.02, 1440) for _ in range(size)], + "population_served": [ + random.choice(["yes", "no"]) for _ in range(size) + ], + "industrial_input": [random.uniform(0, 75) for _ in range(size)], + "sewage_travel_time": [random.uniform(0, 72) for _ in range(size)], + "slopes_flowpop_5m": [random.uniform(-1.3, 1.3) for _ in range(size)], + "pvals_flowpop_5m": [random.uniform(0, 1) for _ in range(size)], + "pdc_flowpop_5m": [random.uniform(-95, 1600) for _ in range(size)], + "predsuprexc_flowpop_5m": [ + random.uniform(-19, 129) for _ in range(size) + ], + "alert_flowpop_5m": [ + random.choice(["yes", "no"]) for _ in range(size) + ], + "slopes_flowpop_3m": [random.uniform(-2, 1.8) for _ in range(size)], + "pvals_flowpop_3m": [random.uniform(0, 1) for _ in range(size)], + "pdc_flowpop_3m": [random.uniform(-95, 1600) for _ in range(size)], + "predsuprexc_flowpop_3m": [ + random.uniform(-19, 129) for _ in range(size) + ], + "alert_flowpop_3m": [ + random.choice(["yes", "no"]) for _ in range(size) + ], + "trend_flowpop_5m3m_p05": [ + random.choice(trend_flowpop_opts) for _ in range(size) + ], + "trend_flowpop_5m3m_p05_prev": [ + random.choice(trend_flowpop_opts) for _ in range(size) + ], + "slopes_mic_5m": [random.uniform(-1.3, 1.3) for _ in range(size)], + "pvals_mic_5m": [random.uniform(0, 1) for _ in range(size)], + "pdc_mic_5m": [random.uniform(-95, 1600) for _ in range(size)], + "predsuprexc_mic_5m": [random.uniform(-19, 129) for _ in range(size)], + "alert_mic_5m": [random.choice(["yes", "no"]) for _ in range(size)], + "slopes_mic_3m": [random.uniform(-2, 1.8) for _ in range(size)], + "pvals_mic_3m": [random.uniform(0, 1) for _ in range(size)], + "pdc_mic_3m": [random.uniform(-95, 1600) for _ in range(size)], + "predsuprexc_mic_3m": [random.uniform(-19, 129) for _ in range(size)], + "alert_mic_3m": [random.choice(["yes", "no"]) for _ in range(size)], + "trend_mic_5m3m_p05": [ + random.choice(trend_flowpop_opts) for _ in range(size) + ], + "trend_mic_5m3m_p05_prev": [ + random.choice(trend_flowpop_opts) for _ in range(size) + ], + "slopes_flowpop_15d": [random.uniform(-1.3, 1.3) for _ in range(size)], + "pvals_flowpop_15d": [random.uniform(0, 1) for _ in range(size)], + "pdc_flowpop_15d": [random.uniform(-95, 1600) for _ in range(size)], + "predsuprexc_flowpop_15d": [ + random.uniform(-19, 129) for _ in range(size) + ], + "alert_flowpop_15d": [ + random.choice(["yes", "no"]) for _ in range(size) + ], + "slopes_flowpop_8d": [random.uniform(-1.3, 1.3) for _ in range(size)], + "pvals_flowpop_8d": [random.uniform(0, 1) for _ in range(size)], + "pdc_flowpop_8d": [random.uniform(-95, 1600) for _ in range(size)], + "predsuprexc_flowpop_8d": [ + random.uniform(-19, 129) for _ in range(size) + ], + "alert_flowpop_8d": [ + random.choice(["yes", "no"]) for _ in range(size) + ], + "trend_flowpop_15d8d_p05": [ + random.choice(trend_flowpop_opts) for _ in range(size) + ], + "trend_flowpop_15d8d_p05_prev": [ + random.choice(trend_flowpop_opts) for _ in range(size) + ], + "slopes_mic_15d": [random.uniform(-1.3, 1.3) for _ in range(size)], + "pvals_mic_15d": [random.uniform(0, 1) for _ in range(size)], + "pdc_mic_15d": [random.uniform(-95, 1600) for _ in range(size)], + "predsuprexc_mic_15d": [random.uniform(-19, 129) for _ in range(size)], + "alert_mic_15d": [random.choice(["yes", "no"]) for _ in range(size)], + "slopes_mic_8d": [random.uniform(-1.3, 1.3) for _ in range(size)], + "pvals_mic_8d": [random.uniform(0, 1) for _ in range(size)], + "pdc_mic_8d": [random.uniform(-95, 1600) for _ in range(size)], + "predsuprexc_mic_8d": [random.uniform(-19, 129) for _ in range(size)], + "alert_mic_8d": [random.choice(["yes", "no"]) for _ in range(size)], + "trend_mic_15d8d_p05": [ + random.choice(trend_flowpop_opts) for _ in range(size) + ], + "trend_mic_15d8d_p05_prev": [ + random.choice(trend_flowpop_opts) for _ in range(size) + ], + "cases_new_cens": [random.randint(0, 100) for _ in range(size)], + "pcr_target_mic_lin_roll7": [ + random.uniform(0, 1) for _ in range(size) + ], + "pcr_target_flowpop_lin_roll7": [ + random.uniform(0, 1) for _ in range(size) + ], + "pcr_target_flowpop_lin_spline": [ + random.uniform(0, 1) for _ in range(size) + ], + "pcr_target_mic_lin_spline": [ + random.uniform(0, 1) for _ in range(size) + ], + "cases_new_cens_per100k_spline": [ + random.uniform(0, 1) for _ in range(size) + ], + "pcr_target_detect": [ + random.choice(["Yes", "No"]) for _ in range(size) + ], + "hum_frac_mic_conc": [random.uniform(0, 1) for _ in range(size)], + "hum_frac_mic_unit": [ + random.choice(["mg/L", "µg/L"]) for _ in range(size) + ], + "hum_frac_chem_conc": [random.uniform(0, 1) for _ in range(size)], + "hum_frac_chem_unit": [ + random.choice(["mg/L", "µg/L"]) for _ in range(size) + ], + "other_norm_conc": [random.uniform(0, 1) for _ in range(size)], + "other_norm_unit": [ + random.choice(["mg/L", "µg/L"]) for _ in range(size) + ], + "lod_sewage": [random.uniform(0, 1) for _ in range(size)], + "pcr_target": [ + random.choice(["sars-cov-2", "omicron", "delta"]) + for _ in range(size) + ], + "pcr_gene_target_agg": [ + random.choice( + [ + "n2", + "cdc n1 (gt-digital)", + "n2 and n1", + "n", + "n1", + "n2 and n and n1", + "n3", + ] + ) + for _ in range(size) + ], + "inhibition_detect": [ + random.choice(["Yes", "No", "Not Tested"]) for _ in range(size) + ], + "inhibition_adjust": [ + random.choice(["Yes", "No"]) for _ in range(size) + ], + "hum_frac_target_chem": [random.uniform(0, 1) for _ in range(size)], + "other_norm_name": [ + random.choice(["name1", "name2"]) for _ in range(size) + ], + "hum_frac_target_mic": [ + random.choice( + [ + "pepper mild mottle virus", + "pmmov (gt-digital)", + ".", + "f+ rna coliphage", + "crassphage", + "hf183", + "pmmov", + "#value!", + ] + ) + for _ in range(size) + ], + "flow_rate": [random.uniform(0, 1) for _ in range(size)], + "pcr_target_avg_conc": [random.uniform(0, 1) for _ in range(size)], + "pcr_target_below_lod": [ + random.choice(["Yes", "No"]) for _ in range(size) + ], + "solids_separation": [ + random.choice(["Yes", "No"]) for _ in range(size) + ], + "concentration_method": [ + random.choice( + [ + "ceres nanotrap", + "innovaprep ultrafiltration", + "membrane filtration with no amendment", + ] + ) + for _ in range(size) + ], + "extraction_method": [ + random.choice( + [ + "nuclisens manual magnetic bead extraction kit", + "qiagen qiaamp buffers with epoch columns", + ] + ) + for _ in range(size) + ], + "rec_eff_target_name": [ + random.choice( + ["heat inactivated sars-cov-2 virus", "puro", "ms2 coliphag"] + ) + for _ in range(size) + ], + "rec_eff_spike_matrix": [ + random.choice(["matrix1", "matrix2"]) for _ in range(size) + ], + "pasteurized": [random.choice(["yes", "no"]) for _ in range(size)], + "pcr_type": [random.choice(["type1", "type2"]) for _ in range(size)], + "quant_stan_type": [ + random.choice(["type1", "type2"]) for _ in range(size) + ], + "pcr_target_units": [ + random.choice(["unit1", "unit2"]) for _ in range(size) + ], + "quality_flag": [random.choice(["yes", "no"]) for _ in range(size)], + "collection_storage_temp": [random.uniform(0, 1) for _ in range(size)], + "collection_storage_time": [random.uniform(0, 1) for _ in range(size)], + "pre_conc_storage_temp": [random.uniform(0, 1) for _ in range(size)], + "pre_conc_storage_time": [random.uniform(0, 1) for _ in range(size)], + "pre_ext_storage_temp": [random.uniform(0, 1) for _ in range(size)], + "pre_ext_storage_time": [random.uniform(0, 1) for _ in range(size)], + "sample_collect_time": [random.uniform(0, 1) for _ in range(size)], + "time_zone": [ + random.choice( + [ + "utc-06:00", + "utc-07:00", + "utc-07:01", + "utc-08:00", + "utc-9", + "utc-8", + "utc-6", + "utc-5", + ] + ) + for _ in range(size) + ], + "collection_water_temp": [random.uniform(0, 1) for _ in range(size)], + "tot_conc_vol": [random.uniform(0, 1) for _ in range(size)], + "ext_blank": [random.uniform(0, 1) for _ in range(size)], + "num_no_target_control": [random.randint(0, 10) for _ in range(size)], + "tss": [random.uniform(0, 1) for _ in range(size)], + "ph": [random.uniform(0, 14) for _ in range(size)], + "composite_freq": [random.uniform(0, 100) for _ in range(size)], + "pcr_target_avg_conc_log10": [ + random.uniform(0, 1) for _ in range(size) + ], + "pcr_target_avg_conc_lin": [random.uniform(0, 1) for _ in range(size)], + "pcr_target_mic_lin": [random.uniform(0, 1) for _ in range(size)], + "pcr_target_mic_log10": [random.uniform(0, 1) for _ in range(size)], + "rec_eff_spike_conc": [random.uniform(0, 1) for _ in range(size)], + "conductivity": [random.uniform(0, 1) for _ in range(size)], + "equiv_sewage_amt": [random.uniform(0, 1) for _ in range(size)], + "hum_frac_mic_conc_lin": [random.uniform(0, 1) for _ in range(size)], + "ntc_amplify": [random.choice(["yes", "no"]) for _ in range(size)], + "ptc_flowpop_3m": [random.uniform(0, 1) for _ in range(size)], + "ptc_flowpop_5m": [random.uniform(0, 1) for _ in range(size)], + "ptc_flowpop_8d": [random.uniform(0, 1) for _ in range(size)], + "ptc_flowpop_15d": [random.uniform(0, 1) for _ in range(size)], + "ptc_mic_3m": [random.uniform(0, 1) for _ in range(size)], + "ptc_mic_5m": [random.uniform(0, 1) for _ in range(size)], + "ptc_mic_8d": [random.uniform(0, 1) for _ in range(size)], + "ptc_mic_15d": [random.uniform(0, 1) for _ in range(size)], + "perc_flowpop": [random.uniform(0, 1) for _ in range(size)], + "quin_flowpop_bin": [ + random.choice(["yes", "no"]) for _ in range(size) + ], + "perc_mic": [random.uniform(0, 1) for _ in range(size)], + "quin_mic_bin": [random.choice(["yes", "no"]) for _ in range(size)], + "major_lab_method": [random.choice(["1", "5"]) for _ in range(size)], + "site_id": [random.randint(1, 100) for _ in range(size)], + "jurisdiction_policy_rid": [fake.uuid4() for _ in range(size)], + } + df = pd.DataFrame(data) + return df if output == "pandas" or output == "pd" else pl.from_pandas(df) diff --git a/cfa/catalog/public/datasets/stf/schemas/param_estimates.py b/cfa/catalog/public/datasets/stf/schemas/param_estimates.py new file mode 100644 index 0000000..a001455 --- /dev/null +++ b/cfa/catalog/public/datasets/stf/schemas/param_estimates.py @@ -0,0 +1,124 @@ +import random + +import pandas as pd +import pandera.pandas as pa +import polars as pl + +geo_list = [ + "AK", + "AL", + "AR", + "AZ", + "CA", + "CO", + "CT", + "DC", + "DE", + "FL", + "GA", + "GU", + "HI", + "IA", + "ID", + "IL", + "IN", + "KS", + "KY", + "LA", + "MA", + "MD", + "ME", + "MI", + "MN", + "MO", + "MS", + "MT", + "NC", + "ND", + "NE", + "NH", + "NJ", + "NM", + "NV", + "NY", + "OH", + "OK", + "OR", + "PA", + "RI", + "SC", + "SD", + "TN", + "TX", + "US", + "UT", + "VA", + "VT", + "WA", + "WI", + "WV", + "WY", +] + +load_schema = pa.DataFrameSchema( + columns={ + "id": pa.Column("int32", pa.Check(lambda x: x >= 3)), + "start_date": pa.Column( + object, + pa.Check(lambda x: pd.to_datetime(x, errors="coerce").notna()), + nullable=True, + coerce=True, + ), # Check if start_date can be converted to datetime + "end_date": pa.Column( + object, + pa.Check(lambda x: pd.to_datetime(x, errors="coerce").notna()), + nullable=True, + coerce=True, + ), # Check if end_date can be converted to datetime + "disease": pa.Column( + str, pa.Check.isin(["COVID-19", "Influenza", "RSV"]) + ), + "parameter": pa.Column( + str, + pa.Check.isin( + ["delay", "right_truncation", "generation_interval"] + ), + ), + "format": pa.Column(str, pa.Check.equal_to("PMF")), + "geo_value": pa.Column(str, pa.Check.isin(geo_list), nullable=True), + "reference_date": pa.Column( + object, + pa.Check(lambda x: pd.to_datetime(x, errors="coerce").notna()), + nullable=True, + coerce=True, + ), + "value": pa.Column( + object, + pa.Check( + lambda x: x.apply( + lambda arr: all(0 <= i <= 1 for i in arr) + and sum(arr) <= 1.01 + ) + ), + nullable=True, + ), # Check if value is a list of numbers between 0 and 1 + } +) + + +def load_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: + df = pd.DataFrame() + df = df.assign( + id=range(3, 3 + size), + start_date=pd.date_range(start="2023-07-01", periods=size), + end_date=pd.date_range(start="2024-07-01", periods=size), + disease=random.choices(["COVID-19", "Influenza", "RSV"], k=size), + parameter=random.choices( + ["delay", "right_truncation", "generation_interval"], k=size + ), + format=["PMF"] * size, + geo_value=random.choices(geo_list, k=size), + reference_date=pd.date_range(start="2023-07-01", periods=size), + value=[[0.1, 0.2, 0.3, 0.4]] * size, + ) + return df if output == "pandas" or output == "pd" else pl.from_pandas(df) diff --git a/cfa/catalog/public/workflows/etl/stf/comprehensive_nssp_gold_deprecated.py b/cfa/catalog/public/workflows/etl/stf/comprehensive_nssp_gold_deprecated.py deleted file mode 100644 index 42be86d..0000000 --- a/cfa/catalog/public/workflows/etl/stf/comprehensive_nssp_gold_deprecated.py +++ /dev/null @@ -1,53 +0,0 @@ -from datetime import datetime -from types import SimpleNamespace - -import polars as pl -from tqdm import tqdm - -from cfa.cloudops.blob_helpers import read_blob_stream -from cfa.dataops import datacat - -from .utils.comp_nssp_version_utils import ( - clear_azure_credentials, - get_all_gold_dates, - get_gold_dates_before_ref, - get_latest_comprehensive_for_date, -) - -dataset = datacat.public.stf.comprehensive_nssp_gold -source_blob = SimpleNamespace(**dataset.config["source"]["storage_location"]) -file_to_copy = "latest_comprehensive.parquet" - - -def copy_file(df: pl.DataFrame | None, date: str | None = None) -> None: - if df is None: - df = pl.read_parquet( - read_blob_stream( - account_name=source_blob.account, - container_name=source_blob.container, - blob_url=file_to_copy, - ).content_as_bytes() - ) - if date is None: - date = datetime.now().isoformat().split("T")[0] - dataset.load.save_dataframe( - df, - path_after_prefix=f"{date}/data.parquet", - ) - - -# this function can backfill the comprehensive dataset based on all nssp gold dates -def generate_versioned_dataset() -> None: - # get all available gold dates and existing versions in the data catalog - versions = dataset.load.get_versions() - gold_dates_sort = get_all_gold_dates() - # get final list to run - date_list = [date for date in gold_dates_sort if date not in versions] - # get data for missing versions and upload - for ref_date in tqdm(date_list): - print(f"Generating comprehensive dataset for {ref_date}...") - dates_available = get_gold_dates_before_ref(ref_date, gold_dates_sort) - df = get_latest_comprehensive_for_date(dates_available) - clear_azure_credentials() - copy_file(df, ref_date) - return None diff --git a/cfa/catalog/public/workflows/etl/stf/nssp_gold_deprecated.py b/cfa/catalog/public/workflows/etl/stf/nssp_gold_deprecated.py index 84a3afc..4f25869 100644 --- a/cfa/catalog/public/workflows/etl/stf/nssp_gold_deprecated.py +++ b/cfa/catalog/public/workflows/etl/stf/nssp_gold_deprecated.py @@ -6,7 +6,7 @@ from cfa.cloudops.blob_helpers import read_blob_stream, walk_blobs_in_container from cfa.dataops import datacat -dataset = datacat.public.stf.nssp_gold +dataset = datacat.public.stf.nssp_gold_v1 source_blob = SimpleNamespace(**dataset.config["source"]["storage_location"]) diff --git a/docs/schemas_and_mock_data.md b/docs/schemas_and_mock_data.md new file mode 100644 index 0000000..9c2c572 --- /dev/null +++ b/docs/schemas_and_mock_data.md @@ -0,0 +1,51 @@ +# Schemas and Mock Data + +## Setting Up Schemas + +To make use of schemas (and mock data) in your workflows, you will need a python file with the name matching the dataset. This file will live within a team's `schema` folder in the catalog. For example, suppose we are working with the nhsn_hrd dataset from the stf team that is defined in `datasets/stf/nhsn_hrd.toml`. The path for the associated schema file would be `datasets/stf/schemas/nhsn_hrd.py`. + +The schema definitions are built on the pandera.pandas package. it is recommended to import it as `import pandera.pandas as pa`. Then one schema can be created for each stage of data (like extract or load). It's recommended to use the names extract_schema or load_schema for consistency, but any name can be used. Each schema should be created by using `pa.DataFrameSchema()`. See the stf/schemas/nssp_gold_v1.py file for an example. + +## Using a Schema + +Now that a schema is defined we can use it in code. We can import our new in the following way: `from cfa.catalog.public.datasets..schemas import `. For our specific example we import it as `from cfa.catalog.public.datasets.stf.schemas import nhsn_hrd`. We can then referenced the schemas here like `schema = nhsn_hrd.load_schema`. + +Schemas are helpful for ensuring data aligns with expected data columns and column types. Once we have a schema and a dataframe (in pandas format) we can validate the dataframe adheres to the schema in the following way: +``` +schema.validate(df) +``` + +## Setting Up Mock Data + +In the same file as the schema for a dataset, we define how our mock data should be generated. It will be in the form of a function and use either `extract_mock_data` or `load_mock_data` as the name, with the following format: +``` +def load_mock_data(output="pandas", size=10) -> pd.DataFrame | pl.DataFrame: + df = pd.DataFrame() + df = df.assign( + --- + ) + return df if output == "pandas" or output == "pd" else pl.from_pandas(df) +``` + +The middle portion of df.assign will be the place to define column names and data generation. This can include random number generation, choosing from options, etc. The general structure for each column is +``` +column_name = [random.randint(0,10) for _ in range(size)], +``` + +Note that column_name is not in quotes. + + +## Using Mock Data + +If datasets have the load_mock_data or extract_mock_data functions defined in their respective schema areas, we can pull this in automatically with cfa-dataops datacat. The mock data functions are accessible at your dataset's load or extract reference endpoint and called `mock_data()`, for example: +``` +ref = datacat.public.stf.nhsn_hrd + +extract_df = ref.extract.mock_data() +load_df = ref.load.mock_data() +``` + +Because of the way we define `extract_mock_data()` and `load_mock_data()` in our schema file, it accepts an `output` parameter of `pandas` or `polars`, and a `size` parameter which is the number of rows in the mock dataframe to generate. For the example above, we can produce a mock dataframe for our load data that is in polars format and 1,000 rows. +``` +df = red.load.mock_data(output = "polars", size = 1000) +``` \ No newline at end of file diff --git a/pyproject.toml b/pyproject.toml index b944b04..1c05884 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -26,7 +26,6 @@ dependencies = [ "faker (>=37.8.0)", "PyGithub (>=2.8.1)" ] - [tool.setuptools.packages.find] namespaces = true include = ["cfa*"]