enh(bokeh): Inspect timeseries

[hoxbro]

Disabling the overlay_aggregate fastpath when we have a selector. A bit of formatting in the applies as it was pretty complicated already.

Code

import itertools

import datashader as ds
import numpy as np
import pandas as pd
import panel as pn

import holoviews as hv
from holoviews.operation.datashader import datashade, dynspread, rasterize

hv.extension("bokeh")

num = 10000
seed = np.random.default_rng(1)


def y(val):
    y = seed.normal(0, 1, num).cumsum()
    if val == 1:
        y[num // 2 : num // 2 + num // 10] = np.nan
    return y + val * 100  # No overlap


dists = [
    pd.DataFrame(
        {
            "x": np.arange(num),
            "y": y(val),
            "val": val,
            "cat": cat,
        }
    )
    for val, cat in [
        (0, "c0"),
        (1, "c1"),
    ]
]

df = pd.concat(dists, ignore_index=True)
curves = hv.Curve(df).groupby("cat").opts(tools=["hover"])
overlay = curves.overlay()


def dynspread_datashade(*args, **kwargs):
    return dynspread(datashade(*args, **kwargs))


def dynspread_rasterize(*args, **kwargs):
    return dynspread(rasterize(*args, **kwargs))


def rasterize_linewidth(*args, **kwargs):
    return rasterize(*args, **dict(kwargs, line_width=10))


def datashade_linewidth(*args, **kwargs):
    return datashade(*args, **dict(kwargs, line_width=10))


ops = (
    rasterize,
    dynspread_rasterize,
    rasterize_linewidth,
    datashade,
    dynspread_datashade,
    datashade_linewidth,
)
aggs = (ds.count(self_intersect=False), ds.by("cat", ds.count(self_intersect=False)))
sels = (None, ds.min("val"))
combinations = itertools.product(ops, aggs, sels)
plots = []
for op, agg, sel in combinations:
    title = f"{op.__name__.replace('_','+')}(agg={type(agg).__name__}, sel={type(sel).__name__ if sel else None})"
    plot = op(overlay, aggregator=agg, selector=sel).opts(
        tools=["hover"], title=title, width=400, height=400
    )
    plots.append(plot)

pn.panel(hv.Layout(plots).cols(4).opts(shared_axes=False)).servable()

[codecov]

Codecov Report

Attention: Patch coverage is 80.00000% with 1 line in your changes missing coverage. Please review.

Project coverage is 88.79%. Comparing base (48f91a5) to head (4eda559).

Files with missing lines	Patch %	Lines
holoviews/operation/datashader.py	80.00%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           inspect    #6479      +/-   ##
===========================================
- Coverage    88.79%   88.79%   -0.01%     
===========================================
  Files          323      323              
  Lines        68885    68887       +2     
===========================================
+ Hits         61166    61167       +1     
- Misses        7719     7720       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[hoxbro]

This is because it would give different colors for a single category. I am not sure if it's worth it, to be honest, but I was a bit confused by the different colors with and without a selector.

Code

import datashader as ds
import numpy as np
import pandas as pd

import holoviews as hv
from holoviews.operation.datashader import datashade, dynspread, rasterize

hv.extension("bokeh")

num = 10000
seed = np.random.default_rng(1)

df = pd.DataFrame({"x": np.arange(num), "y": seed.normal(0, 1, num).cumsum(), "val": 1, "cat": "c0"})
curve = hv.Curve(df)

agg_fn = ds.by("cat", ds.count(self_intersect=False))

p0 = datashade(curve, aggregator=agg_fn, line_width=10, selector=None)
p1 = datashade(curve, aggregator=agg_fn, line_width=10, selector=ds.min("val"))
p0 + p1

Before	After