tabmat as tm, from_csc, renamed bench_glum.py (#448)

* tabmat as tm, from_csc, renamed bench_glum.py

* changelog

* linting (mostly isort because of the new names)

* make mypy ignore tests and fix a benchmark mypy issue

* Remove remnant GCP stuff. Update documentation links.

* Remove deployment and docs actions and workflow steps.

* Required tabmat>=3.0.1

* Remove documentation build from CI.

* Remove old phrase mentioning conda channel.

Co-authored-by: Ben Thompson <[email protected]>

.github/workflows/ci.yml

@@ -34,41 +34,10 @@ jobs:
        uses: actions/[email protected]
        with:
          ref: ${{ github.head_ref }}
-     - uses: google-github-actions/[email protected]
-       with:
-         version: '349.0.0'
-         project_id: ${{ secrets.GCP_UPLOAD_PROJECT_ID }}
-         service_account_email: ${{ secrets.GCP_UPLOAD_SA_EMAIL }}
-         service_account_key: ${{ secrets.GCP_UPLOAD_SA_KEY }}
-         export_default_credentials: true
      - name: Run CI inside of container
        uses: ./.github/actions/unittests
        with:
          python_version: ${{ matrix.PYTHON_VERSION }}
-  docs:
-    name: "Documentation"
-    runs-on: ubuntu-latest
-    env:
-      CI: True
-    strategy:
-      fail-fast: false
-      matrix:
-        PYTHON_VERSION: ['3.9']
-    steps:
-     - name: Pull image
-       run: docker pull condaforge/mambaforge:latest
-     - name: Checkout branch
-       uses: actions/[email protected]
-       with:
-         ref: ${{ github.head_ref }}
-     - uses: google-github-actions/[email protected]
-       with:
-         version: '349.0.0'
-         export_default_credentials: true
-     - name: Run CI inside of container
-       uses: ./.github/actions/docs
-       with:
-         python_version: ${{ matrix.PYTHON_VERSION }}
   linux-conda-build:
     name: "Linux - conda-build - Python ${{ matrix.CONDA_BUILD_YML }}"
     runs-on: ubuntu-latest

.pre-commit-config.yaml

@@ -34,6 +34,7 @@ repos:
        - --check-untyped-defs
        - --ignore-missing-imports
        - --namespace-packages
+      exclude: ^tests/
  - repo: https://github.com/Quantco/pre-commit-mirrors-pyupgrade
    rev: 2.7.2
    hooks:

CHANGELOG.rst

@@ -37,6 +37,7 @@ Unreleased
 **Bug fix:**
 
 - The :meth:`score` method of :class:`~glum.GeneralizedLinearRegressor` and :class:`~glum.GeneralizedLinearRegressorCV` now accepts data frames.
+- Upgrading the code to use tabmat 3.0.0.
 
 **Other:**
 

CONTRIBUTING.md

@@ -1,3 +1,3 @@
 # Contributing
 
-See the [contributing and development information in our documentation](https://docs.dev.***REMOVED***/***REMOVED***/Quantco/glum/latest/contributing.html).
+See the [contributing and development information in our documentation](https://glum.readthedocs.io/en/latest/contributing.html).

README.md

@@ -2,7 +2,7 @@
 
 ![CI](https://github.com/Quantco/glm_benchmarks/workflows/CI/badge.svg)
 
-[Documentation](https://docs.dev.***REMOVED***/***REMOVED***/Quantco/glum/latest/index.html)
+[Documentation](https://glum.readthedocs.io/en/latest/)
 
 Generalized linear models (GLM) are a core statistical tool that include many common methods like least-squares regression, Poisson regression and logistic regression as special cases. At QuantCo, we have used GLMs in e-commerce pricing, insurance claims prediction and more. We have developed `glum`, a fast Python-first GLM library. The development was based on [a fork of scikit-learn](https://github.com/scikit-learn/scikit-learn/pull/9405), so it has a scikit-learn-like API. We are thankful for the starting point provided by Christian Lorentzen in that PR!
 
@@ -19,7 +19,7 @@ This repo also includes tools for benchmarking GLM implementations in the `glum_
 
 ![](docs/_static/headline_benchmark.png)
 
-For more information on `glum`, including tutorials and API reference, please see [the documentation](https://docs.dev.***REMOVED***/***REMOVED***/Quantco/glum/latest/index.html).
+For more information on `glum`, including tutorials and API reference, please see [the documentation](https://glum.readthedocs.io/en/latest/).
 
 # An example: predicting car insurance claim frequency using Poisson regression.
 

conda.recipe/meta.yaml

@@ -41,7 +41,7 @@ requirements:
     - pandas
     - scikit-learn >=0.23
     - scipy
-    - tabmat >=1.0.0
+    - tabmat >=3.0.1
 
 test:
   requires:

docs/contributing.rst

@@ -25,7 +25,7 @@ Pull request process
 Releases
 --------------------------------------------------
 
-- We make package releases infrequently, but usually any time a new non-trivial feature is contributed or a bug is fixed. To make a release, just open a PR that updates the change log with the current date. Once that PR is approved and merged, you can create a new release on [GitHub](https://github.com/Quantco/glum/releases/new). Use the version from the change log as tag and copy the change log entry into the release description. New releases on GitHub are automatically deployed to the QuantCo conda channel.
+- We make package releases infrequently, but usually any time a new non-trivial feature is contributed or a bug is fixed. To make a release, just open a PR that updates the change log with the current date. Once that PR is approved and merged, you can create a new release on [GitHub](https://github.com/Quantco/glum/releases/new). Use the version from the change log as tag and copy the change log entry into the release description. 
 
 Install for development
 --------------------------------------------------

docs/install.rst

@@ -6,4 +6,4 @@ You can install the package through conda:
 
    conda install glum -c conda-forge
 
-Head onwards to :doc:`Getting Started<getting_started/getting_started>` to try it out!
+Head onwards to :doc:`Getting Started<getting_started/getting_started>` to try it out!

docs/tutorials/glm_french_motor_tutorial/glm_french_motor.ipynb

@@ -466,7 +466,7 @@
     "- `alpha_search=True`: tells the GLM to search along the regularization path for the best alpha\r\n",
     "- `l1_ratio = 1` tells the GLM to only use l1 penalty (not l2). `l1_ratio` is the elastic net mixing parameter. For ``l1_ratio = 0``, the penalty is an L2 penalty. ``For l1_ratio = 1``, it is an L1 penalty.  For ``0 < l1_ratio < 1``, the penalty is a combination of L1 and L2.\r\n",
     "\r\n",
-    "See the `GeneralizedLinearRegressor` class [API documentation](https://docs.dev.***REMOVED***/***REMOVED***/Quantco/glum/latest/api/modules.html) for more details.\r\n",
+    "See the `GeneralizedLinearRegressor` class [API documentation](https://glum.readthedocs.io/en/latest/api/modules.html) for more details.\r\n",
     "\r\n",
     "*Note*: `glum` also supported a cross validation model GeneralizedLinearRegressorCV. However, because cross validation requires fitting many models, it is much slower and we don’t demonstrate it in this tutorial."
    ],
@@ -1305,4 +1305,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 4
-}
+}

docs/tutorials/glm_french_motor_tutorial/glm_french_motor.md

@@ -159,7 +159,7 @@ Now, we define our GLM using the `GeneralizedLinearRegressor` class from `glum`.
 - `alpha_search=True`: tells the GLM to search along the regularization path for the best alpha
 - `l1_ratio = 1` tells the GLM to only use l1 penalty (not l2). `l1_ratio` is the elastic net mixing parameter. For ``l1_ratio = 0``, the penalty is an L2 penalty. ``For l1_ratio = 1``, it is an L1 penalty.  For ``0 < l1_ratio < 1``, the penalty is a combination of L1 and L2.
 
-See the `GeneralizedLinearRegressor` class [API documentation](https://docs.dev.***REMOVED***/***REMOVED***/Quantco/glum/latest/api/modules.html) for more details.
+See the `GeneralizedLinearRegressor` class [API documentation](https://glum.readthedocs.io/en/latest/api/modules.html) for more details.
 
 *Note*: `glum` also supported a cross validation model GeneralizedLinearRegressorCV. However, because cross validation requires fitting many models, it is much slower and we don’t demonstrate it in this tutorial.
 

environment.yml

@@ -8,7 +8,7 @@ dependencies:
   - libblas>=0=*mkl
   - numexpr
   - pandas
-  - tabmat>=3.0.0
+  - tabmat>=3.0.1
   - scikit-learn >= 0.23
   - scipy
   - tqdm

src/glum/_distribution.py

@@ -4,8 +4,8 @@
 
 import numexpr
 import numpy as np
-from tabmat import MatrixBase, StandardizedMatrix
 from scipy import sparse, special
+from tabmat import MatrixBase, StandardizedMatrix
 
 from ._functions import (
     binomial_logit_eta_mu_deviance,