scikit-learn · adrinjalali · Sep 23, 2021 · Sep 7, 2021 · Sep 8, 2021 · Sep 8, 2021
diff --git a/.circleci/config.yml b/.circleci/config.yml
@@ -109,27 +109,6 @@ jobs:
           name: linting
           command: ./build_tools/circle/linting.sh
 
-  pypy3:
-    docker:
-      - image: condaforge/miniforge3
-    environment:
-      # Avoid the interactive dialog when installing tzdata
-      - DEBIAN_FRONTEND: noninteractive
-    steps:
-      - restore_cache:
-          keys:
-            - pypy3-ccache-{{ .Branch }}
-            - pypy3-ccache
-      - run: apt-get -yq update && apt-get -yq install git ssh
-      - checkout
-      - run: conda init bash && source ~/.bashrc
-      - run: ./build_tools/circle/build_test_pypy.sh
-      - save_cache:
-          key: pypy3-ccache-{{ .Branch }}-{{ .BuildNum }}
-          paths:
-            - ~/.ccache
-            - ~/.cache/pip
-
   linux-arm64:
     machine:
       image: ubuntu-2004:202101-01
@@ -190,16 +169,6 @@ workflows:
       - deploy:
           requires:
             - doc
-  pypy:
-    triggers:
-      - schedule:
-          cron: "0 0 * * *"
-          filters:
-            branches:
-              only:
-                - main
-    jobs:
-      - pypy3
   linux-arm64:
     jobs:
       - linux-arm64
diff --git a/azure-pipelines.yml b/azure-pipelines.yml
@@ -182,9 +182,9 @@ jobs:
         TEST_DOCSTRINGS: 'true'
         CHECK_WARNINGS: 'true'
 
-- template: build_tools/azure/posix-32.yml
+- template: build_tools/azure/posix-docker.yml
   parameters:
-    name: Linux32
+    name: Linux_Docker
     vmImage: ubuntu-20.04
     dependsOn: [linting, git_commit]
     condition: |
@@ -194,8 +194,14 @@ jobs:
         ne(variables['Build.Reason'], 'Schedule')
       )
     matrix:
+      pypy3:
+        DISTRIB: 'conda-mamba-pypy3'
+        DOCKER_CONTAINER: 'condaforge/mambaforge-pypy3:4.10.3-5'
+        PILLOW_VERSION: 'none'
+        PANDAS_VERSION: 'none'
       debian_atlas_32bit:
         DISTRIB: 'debian-32'
+        DOCKER_CONTAINER: 'i386/debian:10.9'
         JOBLIB_VERSION: 'min'
         # disable pytest xdist due to unknown bug with 32-bit container
         PYTEST_XDIST_VERSION: 'none'

diff --git a/benchmarks/bench_mnist.py b/benchmarks/bench_mnist.py
@@ -6,7 +6,7 @@
 Benchmark on the MNIST dataset.  The dataset comprises 70,000 samples
 and 784 features. Here, we consider the task of predicting
 10 classes -  digits from 0 to 9 from their raw images. By contrast to the
-covertype dataset, the feature space is homogenous.
+covertype dataset, the feature space is homogeneous.
 
 Example of output :
     [..]

diff --git a/benchmarks/bench_random_projections.py b/benchmarks/bench_random_projections.py
@@ -43,10 +43,10 @@ def compute_time(t_start, delta):
     return delta.seconds + delta.microseconds / mu_second
 
 
-def bench_scikit_transformer(X, transfomer):
+def bench_scikit_transformer(X, transformer):
     gc.collect()
 
-    clf = clone(transfomer)
+    clf = clone(transformer)
 
     # start time
     t_start = datetime.now()
@@ -195,7 +195,7 @@ def print_row(clf_type, time_fit, time_transform):
     ###########################################################################
     n_nonzeros = int(opts.ratio_nonzeros * opts.n_features)
 
-    print("Dataset statics")
+    print("Dataset statistics")
     print("===========================")
     print("n_samples \t= %s" % opts.n_samples)
     print("n_features \t= %s" % opts.n_features)

diff --git a/build_tools/azure/install.sh b/build_tools/azure/install.sh
@@ -5,9 +5,19 @@ set -x
 
 UNAMESTR=`uname`
 
+if [[ "$DISTRIB" == "conda-mamba-pypy3" ]]; then
+    # condaforge/mambaforge-pypy3 needs compilers
+    apt-get -yq update
+    apt-get -yq install build-essential
+fi
+
 make_conda() {
     TO_INSTALL="$@"
-    conda create -n $VIRTUALENV --yes $TO_INSTALL
+    if [[ "$DISTRIB" == *"mamba"* ]]; then
+        mamba create -n $VIRTUALENV --yes $TO_INSTALL
+    else
+        conda create -n $VIRTUALENV --yes $TO_INSTALL
+    fi
     source activate $VIRTUALENV
 }
 
@@ -25,15 +35,21 @@ setup_ccache() {
 # imports get_dep
 source build_tools/shared.sh
 
-if [[ "$DISTRIB" == "conda" ]]; then
+if [[ "$DISTRIB" == "conda" || "$DISTRIB" == *"mamba"* ]]; then
 
     if [[ "$CONDA_CHANNEL" != "" ]]; then
         TO_INSTALL="-c $CONDA_CHANNEL"
     else
         TO_INSTALL=""
     fi
 
-    TO_INSTALL="$TO_INSTALL python=$PYTHON_VERSION ccache pip blas[build=$BLAS]"
+    if [[ "$DISTRIB" == *"pypy"* ]]; then
+        TO_INSTALL="$TO_INSTALL pypy"
+    else
+        TO_INSTALL="$TO_INSTALL python=$PYTHON_VERSION"
+    fi
+
+    TO_INSTALL="$TO_INSTALL ccache pip blas[build=$BLAS]"
 
     TO_INSTALL="$TO_INSTALL $(get_dep numpy $NUMPY_VERSION)"
     TO_INSTALL="$TO_INSTALL $(get_dep scipy $SCIPY_VERSION)"

diff --git a/build_tools/azure/posix-32.yml → build_tools/azure/posix-docker.yml b/build_tools/azure/posix-32.yml → build_tools/azure/posix-docker.yml
@@ -30,12 +30,16 @@ jobs:
     THREADPOOLCTL_VERSION: 'latest'
     COVERAGE: 'false'
     TEST_DOCSTRINGS: 'false'
+    BLAS: 'openblas'
+    # Set in azure-pipelines.yml
+    DISTRIB: ''
+    DOCKER_CONTAINER: ''
   strategy:
     matrix:
       ${{ insert }}: ${{ parameters.matrix }}
 
   steps:
-    # Container is detached and sleeping, allowing steps to run commmands
+    # Container is detached and sleeping, allowing steps to run commands
     # in the container. The TEST_DIR is mapped allowing the host to access
     # the JUNITXML file
     - script: >
@@ -45,7 +49,7 @@ jobs:
         -w /io
         --detach
         --name skcontainer
-        -e DISTRIB=debian-32
+        -e DISTRIB=$DISTRIB
         -e TEST_DIR=/temp_dir
         -e JUNITXML=$JUNITXML
         -e VIRTUALENV=testvenv
@@ -63,7 +67,8 @@ jobs:
         -e OMP_NUM_THREADS=$OMP_NUM_THREADS
         -e OPENBLAS_NUM_THREADS=$OPENBLAS_NUM_THREADS
         -e SKLEARN_SKIP_NETWORK_TESTS=$SKLEARN_SKIP_NETWORK_TESTS
-        i386/debian:10.9
+        -e BLAS=$BLAS
+        $DOCKER_CONTAINER
         sleep 1000000
       displayName: 'Start container'
     - script: >

diff --git a/build_tools/circle/list_versions.py b/build_tools/circle/list_versions.py
@@ -34,7 +34,7 @@ def human_readable_data_quantity(quantity, multiple=1024):
 
 def get_file_extension(version):
     if "dev" in version:
-        # The 'dev' branch should be explictly handled
+        # The 'dev' branch should be explicitly handled
         return "zip"
 
     current_version = LooseVersion(version)

diff --git a/build_tools/shared.sh b/build_tools/shared.sh
@@ -5,7 +5,7 @@ get_dep() {
         # do not install with none
         echo
     elif [[ "${version%%[^0-9.]*}" ]]; then
-        # version number is explicity passed
+        # version number is explicitly passed
         echo "$package==$version"
     elif [[ "$version" == "latest" ]]; then
         # use latest

diff --git a/doc/common_pitfalls.rst b/doc/common_pitfalls.rst
@@ -560,7 +560,7 @@ bad performance. Similarly, we want a random forest to be robust w.r.t the
 set of randomly selected features that each tree will be using.
 
 For these reasons, it is preferable to evaluate the cross-validation
-preformance by letting the estimator use a different RNG on each fold. This
+performance by letting the estimator use a different RNG on each fold. This
 is done by passing a `RandomState` instance (or `None`) to the estimator
 initialization.
 

diff --git a/doc/conf.py b/doc/conf.py
@@ -240,7 +240,7 @@
     "release_highlights"
 ] = f"auto_examples/release_highlights/{latest_highlights}"
 
-# get version from higlight name assuming highlights have the form
+# get version from highlight name assuming highlights have the form
 # plot_release_highlights_0_22_0
 highlight_version = ".".join(latest_highlights.split("_")[-3:-1])
 html_context["release_highlights_version"] = highlight_version

diff --git a/doc/developers/advanced_installation.rst b/doc/developers/advanced_installation.rst
@@ -60,11 +60,12 @@ feature, code or documentation improvement).
 #. Optional (but recommended): create and activate a dedicated virtualenv_
    or `conda environment`_.
 
-#. Install Cython_ and build the project with pip in :ref:`editable_mode`:
+#. Install NumPy_, SciPy_, and Cython_ and build the project with pip in
+   :ref:`editable_mode`:
 
    .. prompt:: bash $
 
-        pip install cython
+        pip install numpy scipy cython
         pip install --verbose --no-build-isolation --editable .
 
 #. Check that the installed scikit-learn has a version number ending with
@@ -100,9 +101,6 @@ runtime:
 - Joblib (>= |JoblibMinVersion|),
 - threadpoolctl (>= |ThreadpoolctlMinVersion|).
 
-Those dependencies are **automatically installed by pip** if they were missing
-when building scikit-learn from source.
-
 .. note::
 
    For running on PyPy, PyPy3-v5.10+, Numpy 1.14.0+, and scipy 1.1.0+
@@ -376,7 +374,7 @@ isolation from the Python packages installed via the system packager. When
 using an isolated environment, ``pip3`` should be replaced by ``pip`` in the
 above commands.
 
-When precompiled wheels of the runtime dependencies are not avalaible for your
+When precompiled wheels of the runtime dependencies are not available for your
 architecture (e.g. ARM), you can install the system versions:
 
 .. prompt:: bash $
@@ -436,6 +434,8 @@ the base system and these steps will not be necessary.
 
 .. _OpenMP: https://en.wikipedia.org/wiki/OpenMP
 .. _Cython: https://cython.org
+.. _NumPy: https://numpy.org
+.. _SciPy: https://www.scipy.org
 .. _Homebrew: https://brew.sh
 .. _virtualenv: https://docs.python.org/3/tutorial/venv.html
 .. _conda environment: https://docs.conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html

diff --git a/doc/developers/contributing.rst b/doc/developers/contributing.rst
@@ -1004,7 +1004,7 @@ installed in your current Python environment:
 
   asv run --python=same
 
-It's particulary useful when you installed scikit-learn in editable mode to
+It's particularly useful when you installed scikit-learn in editable mode to
 avoid creating a new environment each time you run the benchmarks. By default
 the results are not saved when using an existing installation. To save the
 results you must specify a commit hash:

diff --git a/doc/developers/maintainer.rst b/doc/developers/maintainer.rst
@@ -33,7 +33,7 @@ Before a release
 
    - ``maint_tools/sort_whats_new.py`` can put what's new entries into
      sections. It's not perfect, and requires manual checking of the changes.
-     If the whats new list is well curated, it may not be necessary.
+     If the what's new list is well curated, it may not be necessary.
 
    - The ``maint_tools/whats_missing.sh`` script may be used to identify pull
      requests that were merged but likely missing from What's New.
@@ -198,7 +198,7 @@ Making a release
   `Continuous Integration
   <https://en.wikipedia.org/wiki/Continuous_integration>`_. The CD workflow on
   GitHub Actions is also used to automatically create nightly builds and
-  publish packages for the developement branch of scikit-learn. See
+  publish packages for the development branch of scikit-learn. See
   :ref:`install_nightly_builds`.
 
 4. Once all the CD jobs have completed successfully in the PR, merge it,

diff --git a/doc/install.rst b/doc/install.rst
@@ -158,7 +158,7 @@ Installing on Apple Silicon M1 hardware
 
 The recently introduced `macos/arm64` platform (sometimes also known as
 `macos/aarch64`) requires the open source community to upgrade the build
-configuation and automation to properly support it.
+configuration and automation to properly support it.
 
 At the time of writing (January 2021), the only way to get a working
 installation of scikit-learn on this hardware is to install scikit-learn and its
@@ -204,7 +204,7 @@ It can be installed by typing the following command:
 Debian/Ubuntu
 -------------
 
-The Debian/Ubuntu package is splitted in three different packages called
+The Debian/Ubuntu package is split in three different packages called
 ``python3-sklearn`` (python modules), ``python3-sklearn-lib`` (low-level
 implementations and bindings), ``python3-sklearn-doc`` (documentation).
 Only the Python 3 version is available in the Debian Buster (the more recent

diff --git a/doc/modules/compose.rst b/doc/modules/compose.rst
@@ -449,13 +449,13 @@ By default, the remaining rating columns are ignored (``remainder='drop'``)::
   >>> column_trans = ColumnTransformer(
   ...     [('categories', OneHotEncoder(dtype='int'), ['city']),
   ...      ('title_bow', CountVectorizer(), 'title')],
-  ...     remainder='drop', prefix_feature_names_out=False)
+  ...     remainder='drop', verbose_feature_names_out=False)
 
   >>> column_trans.fit(X)
-  ColumnTransformer(prefix_feature_names_out=False,
-                    transformers=[('categories', OneHotEncoder(dtype='int'),
+  ColumnTransformer(transformers=[('categories', OneHotEncoder(dtype='int'),
                                    ['city']),
-                                  ('title_bow', CountVectorizer(), 'title')])
+                                  ('title_bow', CountVectorizer(), 'title')],
+                    verbose_feature_names_out=False)
 
   >>> column_trans.get_feature_names_out()
   array(['city_London', 'city_Paris', 'city_Sallisaw', 'bow', 'feast',
@@ -573,7 +573,7 @@ many estimators. This visualization is activated by setting the
 
   >>> from sklearn import set_config
   >>> set_config(display='diagram')   # doctest: +SKIP
-  >>> # diplays HTML representation in a jupyter context
+  >>> # displays HTML representation in a jupyter context
   >>> column_trans  # doctest: +SKIP
 
 An example of the HTML output can be seen in the

diff --git a/doc/modules/cross_decomposition.rst b/doc/modules/cross_decomposition.rst
@@ -64,7 +64,7 @@ Set :math:`X_1` to :math:`X` and :math:`Y_1` to :math:`Y`. Then, for each
   :math:`C = X_k^T Y_k`.
   :math:`u_k` and :math:`v_k` are called the *weights*.
   By definition, :math:`u_k` and :math:`v_k` are
-  choosen so that they maximize the covariance between the projected
+  chosen so that they maximize the covariance between the projected
   :math:`X_k` and the projected target, that is :math:`\text{Cov}(X_k u_k,
   Y_k v_k)`.
 - b) Project :math:`X_k` and :math:`Y_k` on the singular vectors to obtain

diff --git a/doc/modules/cross_validation.rst b/doc/modules/cross_validation.rst
@@ -974,7 +974,7 @@ test is therefore only able to show when the model reliably outperforms
 random guessing.
 
 Finally, :func:`~sklearn.model_selection.permutation_test_score` is computed
-using brute force and interally fits ``(n_permutations + 1) * n_cv`` models.
+using brute force and internally fits ``(n_permutations + 1) * n_cv`` models.
 It is therefore only tractable with small datasets for which fitting an
 individual model is very fast.
 

diff --git a/doc/modules/decomposition.rst b/doc/modules/decomposition.rst
@@ -829,7 +829,7 @@ and the intensity of the regularization with the :attr:`alpha_W` and :attr:`alph
 (:math:`\alpha_W` and :math:`\alpha_H`) parameters. The priors are scaled by the number
 of samples (:math:`n\_samples`) for `H` and the number of features (:math:`n\_features`)
 for `W` to keep their impact balanced with respect to one another and to the data fit
-term as independant as possible of the size of the training set. Then the priors terms
+term as independent as possible of the size of the training set. Then the priors terms
 are:
 
 .. math::

diff --git a/doc/modules/lda_qda.rst b/doc/modules/lda_qda.rst
@@ -187,7 +187,7 @@ an estimate for the covariance matrix). Setting this parameter to a value
 between these two extrema will estimate a shrunk version of the covariance
 matrix.
 
-The shrinked Ledoit and Wolf estimator of covariance may not always be the
+The shrunk Ledoit and Wolf estimator of covariance may not always be the
 best choice. For example if the distribution of the data
 is normally distributed, the
 Oracle Shrinkage Approximating estimator :class:`sklearn.covariance.OAS`
@@ -234,7 +234,7 @@ For QDA, the use of the SVD solver relies on the fact that the covariance
 matrix :math:`\Sigma_k` is, by definition, equal to :math:`\frac{1}{n - 1}
 X_k^tX_k = \frac{1}{n - 1} V S^2 V^t` where :math:`V` comes from the SVD of the (centered)
 matrix: :math:`X_k = U S V^t`. It turns out that we can compute the
-log-posterior above without having to explictly compute :math:`\Sigma`:
+log-posterior above without having to explicitly compute :math:`\Sigma`:
 computing :math:`S` and :math:`V` via the SVD of :math:`X` is enough. For
 LDA, two SVDs are computed: the SVD of the centered input matrix :math:`X`
 and the SVD of the class-wise mean vectors.

diff --git a/doc/modules/model_evaluation.rst b/doc/modules/model_evaluation.rst
@@ -2381,7 +2381,7 @@ of 0.0.
 A scorer object with a specific choice of ``power`` can be built by::
 
   >>> from sklearn.metrics import d2_tweedie_score, make_scorer
-  >>> d2_tweedie_score_15 = make_scorer(d2_tweedie_score, pwoer=1.5)
+  >>> d2_tweedie_score_15 = make_scorer(d2_tweedie_score, power=1.5)
 
 .. _pinball_loss: