julearn 0.3.5.dev56__tar.gz → 0.3.5.dev67__tar.gz

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/.github/workflows/check-stale.yml

@@ -8,7 +8,7 @@ jobs:
   stale:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/stale@v9
+      - uses: actions/stale@v10
         with:
           stale-issue-message: 'This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 7 days.'
           days-before-stale: 30

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/.github/workflows/ci.yml

@@ -29,7 +29,7 @@ jobs:
 
     steps:
     - uses: actions/checkout@v5
-    - uses: actions/setup-python@v5
+    - uses: actions/setup-python@v6
       with:
         python-version: ${{ matrix.python-version }}
     - name: Install Python dependencies

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/.github/workflows/docs-preview.yml

@@ -22,7 +22,7 @@ jobs:
     - uses: actions/checkout@v5
       with:
         fetch-depth: 0
-    - uses: actions/setup-python@v5
+    - uses: actions/setup-python@v6
       with:
           python-version: "3.x"
     - name: Install Python dependencies

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/.github/workflows/docs.yml

@@ -18,7 +18,7 @@ jobs:
       with:
         # require all of history to see all tagged versions' docs
         fetch-depth: 0
-    - uses: actions/setup-python@v5
+    - uses: actions/setup-python@v6
       with:
         python-version: "3.x"
     - name: Install Python dependencies
@@ -39,7 +39,7 @@ jobs:
         make -C docs/ html
         touch docs/_build/.nojekyll
         cp docs/redirect.html docs/_build/index.html
-    - uses: peaceiris/actions-gh-pages@v3
+    - uses: peaceiris/actions-gh-pages@v4
       with:
         github_token: ${{ secrets.GITHUB_TOKEN }}
         publish_dir: docs/_build

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/.github/workflows/lint.yml

@@ -9,7 +9,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v5
-      - uses: actions/setup-python@v5
+      - uses: actions/setup-python@v6
         with:
           python-version: "3.x"
       - name: Install Python dependencies

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/.github/workflows/publish.yml

@@ -21,7 +21,7 @@ jobs:
       with:
         fetch-depth: 0
         submodules: true
-    - uses: actions/setup-python@v5
+    - uses: actions/setup-python@v6
       with:
         python-version: "3.x"
     - name: Install build

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: julearn
-Version: 0.3.5.dev56
+Version: 0.3.5.dev67
 Summary: Juelich Machine Learning Library
 Author-email: Fede Raimondo <f.raimondo@fz-juelich.de>, Sami Hamdan <s.hamdan@fz-juelich.de>
 Maintainer-email: Sami Hamdan <s.hamdan@fz-juelich.de>

julearn-0.3.5.dev67/docs/changes/newsfragments/254.bugfix

@@ -0,0 +1 @@
+Fix :class:`.StratifiedBootstrap` behaviour to not leak test and train indices during splitting by `Synchon Mandal`_

julearn-0.3.5.dev67/docs/changes/newsfragments/254.change

@@ -0,0 +1 @@
+Remove ``test_size`` and ``train_size`` parameters and change default value of ``n_splits`` from 5 to 200 for :class:`.StratifiedBootstrap` by `Synchon Mandal`_

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/examples/04_confounds/plot_confound_removal_classification.py

@@ -58,7 +58,7 @@ confounds = ["petal_width"]
 # change the number of bootstrap iterations to at least 2000 for a valid test.
 n_bootstrap = 20
 n_elements = len(df_iris)
-cv = StratifiedBootstrap(n_splits=n_bootstrap, test_size=0.3, random_state=42)
+cv = StratifiedBootstrap(n_splits=n_bootstrap, random_state=42)
 
 ###############################################################################
 # First, we will train a model without performing confound removal on features.
@@ -79,7 +79,7 @@ scores_ncr = run_cross_validation(
 ###############################################################################
 # Next, we train a model after performing confound removal on the features.
 # Note: we initialize the CV again to use the same folds as before.
-cv = StratifiedBootstrap(n_splits=n_bootstrap, test_size=0.3, random_state=42)
+cv = StratifiedBootstrap(n_splits=n_bootstrap, random_state=42)
 
 # In order to tell ``run_cross_validation`` which columns are confounds,
 # and which columns are features, we have to define the X_types:

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/julearn/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.3.5.dev56'
-__version_tuple__ = version_tuple = (0, 3, 5, 'dev56')
+__version__ = version = '0.3.5.dev67'
+__version_tuple__ = version_tuple = (0, 3, 5, 'dev67')
 
-__commit_id__ = commit_id = 'gf35d9bb15'
+__commit_id__ = commit_id = 'geb8b5927e'

julearn-0.3.5.dev67/julearn/model_selection/stratified_bootstrap.py

@@ -0,0 +1,182 @@
+"""Class for Stratified Bootstrap cross-validation iterator."""
+
+# Authors: Federico Raimondo <f.raimondo@fz-juelich.de>
+#          Sami Hamdan <s.hamdan@fz-juelich.de>
+#          Synchon Mandal <s.mandal@fz-juelich.de>
+# License: AGPL
+
+from typing import ClassVar, Optional, Union
+
+import numpy as np
+from numpy.random import RandomState
+from sklearn.model_selection._split import (
+    _build_repr,
+    _UnsupportedGroupCVMixin,
+)
+from sklearn.utils import check_array, metadata_routing
+from sklearn.utils._array_api import _convert_to_numpy, get_namespace
+from sklearn.utils.extmath import _approximate_mode
+from sklearn.utils.metadata_routing import _MetadataRequester
+from sklearn.utils.validation import (
+    _num_samples,
+    check_random_state,
+    indexable,
+)
+
+
+class StratifiedBootstrap(_MetadataRequester, _UnsupportedGroupCVMixin):
+    """Class-wise stratified bootstrap cross-validator.
+
+    Provides train/test indices to split data in train/test sets.
+
+    This cross-validation object returns stratified randomized folds. The folds
+    are made by preserving the percentage of samples for each class in `y` in a
+    binary or multiclass classification setting.
+
+    Parameters
+    ----------
+    n_splits : int, default=200
+        Number of bootstrap iterations.
+    random_state : int or RandomState instance or None, default=None
+        Controls the randomness of the training and testing indices produced.
+        Pass an int for reproducible output across multiple function calls.
+
+    """
+
+    # This indicates that by default CV splitters don't have a "groups" kwarg,
+    # unless indicated by inheriting from ``GroupsConsumerMixin``.
+    # This also prevents ``set_split_request`` to be generated for splitters
+    # which don't support ``groups``.
+    __metadata_request__split: ClassVar = {"groups": metadata_routing.UNUSED}
+
+    def __init__(
+        self,
+        n_splits: int = 200,
+        random_state: Optional[Union[int, RandomState]] = None,
+    ) -> None:
+        self.n_splits = n_splits
+        self.random_state = random_state
+
+    def split(
+        self,
+        X,  # noqa: N803
+        y,
+        groups=None,
+    ):
+        """Generate indices to split data into training and test set.
+
+        Parameters
+        ----------
+        X : array-like of shape (n_samples, n_features)
+            Training data, where `n_samples` is the number of samples
+            and `n_features` is the number of features.
+        y : array-like of shape (n_samples,)
+            The target variable for supervised learning problems.
+        groups : array-like of shape (n_samples,), default=None
+            Group labels for the samples used while splitting the dataset into
+            train/test set.
+
+        Yields
+        ------
+        train : ndarray
+            The training set indices for that split.
+        test : ndarray
+            The testing set indices for that split.
+
+        Notes
+        -----
+        Randomized CV splitters may return different results for each call of
+        split. You can make the results identical by setting `random_state`
+        to an integer.
+
+        """
+        # From BaseCrossValidator
+        X, y, groups = indexable(X, y, groups)
+        # From StratifiedShuffleSplit
+        n_samples = _num_samples(X)
+        y = check_array(y, input_name="y", ensure_2d=False, dtype=None)
+
+        # Convert to numpy as not all operations are supported by the Array
+        # API. `y` is probably never a very large array, which means that
+        # converting it should be cheap
+        xp, _ = get_namespace(y)
+        y = _convert_to_numpy(y, xp=xp)
+
+        if y.ndim == 2:
+            # for multi-label y, map each distinct row to a string repr
+            # using join because str(row) uses an ellipsis if len(row) > 1000
+            y = np.array([" ".join(row.astype("str")) for row in y])
+
+        classes, y_indices = np.unique(y, return_inverse=True)
+        n_classes = classes.shape[0]
+
+        class_counts = np.bincount(y_indices)
+        if np.min(class_counts) < 2:
+            raise ValueError(
+                "The least populated class in y has only 1"
+                " member, which is too few. The minimum"
+                " number of groups for any class cannot"
+                " be less than 2."
+            )
+
+        # Find the sorted list of instances for each class:
+        # (np.unique above performs a sort, so code is O(n logn) already)
+        class_indices = np.split(
+            np.argsort(y_indices, kind="mergesort"),
+            np.cumsum(class_counts)[:-1],
+        )
+
+        rng = check_random_state(self.random_state)
+
+        for _ in range(self.n_splits):
+            # if there are ties in the class-counts, we want
+            # to make sure to break them anew in each iteration
+            n_i = _approximate_mode(class_counts, n_samples, rng)
+
+            train = []
+            test = []
+
+            # Adapted from mlxtend.BootstrapOutOfBag
+            for i in range(n_classes):
+                train_idx = rng.choice(
+                    class_indices[i], size=n_i[i], replace=True
+                )
+                test_idx = np.array(
+                    list(set(class_indices[i]) - set(train_idx))
+                )
+                train.extend(train_idx)
+                test.extend(test_idx)
+
+            train = rng.permutation(train)
+            test = rng.permutation(test)
+
+            yield train, test
+
+    def get_n_splits(
+        self,
+        X=None,  # noqa: N803
+        y=None,
+        groups=None,
+    ) -> int:
+        """Return the number of splitting iterations in the cross-validator.
+
+        Parameters
+        ----------
+        X : object
+            Always ignored, exists for compatibility.
+        y : object
+            Always ignored, exists for compatibility.
+        groups : object
+            Always ignored, exists for compatibility.
+
+        Returns
+        -------
+        n_splits : int
+            Returns the number of splitting iterations in the cross-validator.
+
+        """
+        return self.n_splits  # pragma: no cover
+
+    def __repr__(self) -> str:
+        """Object representation."""
+        return _build_repr(self)  # pragma: no cover

julearn-0.3.5.dev67/julearn/model_selection/tests/test_stratified_bootstrap.py

@@ -0,0 +1,81 @@
+"""Provides tests for the stratified bootstrap CV generator."""
+
+# Authors: Federico Raimondo <f.raimondo@fz-juelich.de>
+#          Synchon Mandal <s.mandal@fz-juelich.de>
+# License: AGPL
+
+import pytest
+from sklearn.datasets import make_classification
+
+from julearn.model_selection import StratifiedBootstrap
+
+
+def test_stratified_bootstrap_error() -> None:
+    """Test stratified bootstrap error."""
+    X, y = make_classification(n_samples=1, n_classes=1)
+    with pytest.raises(
+        ValueError, match="The least populated class in y has only 1"
+    ):
+        cv = StratifiedBootstrap()
+        list(cv.split(X, y))
+
+
+@pytest.mark.parametrize(
+    "n_classes, n_splits",
+    [
+        (2, 50),
+        (2, 100),
+        (2, 200),
+        (3, 50),
+        (3, 100),
+        (3, 200),
+        (4, 50),
+        (4, 100),
+        (4, 200),
+        (5, 50),
+        (5, 100),
+        (5, 200),
+        (6, 50),
+        (6, 100),
+        (6, 200),
+        (7, 50),
+        (7, 100),
+        (7, 200),
+        (8, 50),
+        (8, 100),
+        (8, 200),
+        (9, 50),
+        (9, 100),
+        (9, 200),
+        (10, 50),
+        (10, 100),
+        (10, 200),
+    ],
+)
+def test_stratified_bootstrap(n_classes: int, n_splits: int) -> None:
+    """Test stratified bootstrap CV generator.
+
+    Parameters
+    ----------
+    n_classes : int
+        The parametrized number of classes (or strata).
+    n_splits : int
+        The parametrized number of splits.
+
+    """
+    samples = 100 * n_classes
+    X, y = make_classification(
+        n_samples=samples,
+        n_features=n_classes * 10,
+        n_informative=n_classes,
+        n_redundant=0,
+        n_repeated=0,
+        n_classes=n_classes,
+    )
+    cv = StratifiedBootstrap(n_splits=n_splits, random_state=42)
+    # Check splits
+    results = list(cv.split(X, y))
+    assert len(results) == n_splits
+    # Check train and test indices are disjoint
+    for train_idxs, test_idxs in cv.split(X, y):
+        assert frozenset(train_idxs).isdisjoint(test_idxs)

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/julearn/pipeline/tests/test_merger.py

@@ -97,7 +97,7 @@ def test_merger_errors() -> None:
     pipe3 = GridSearchCV(SVC(), param_grid={"C": [1, 10]})
     with pytest.raises(
         ValueError,
-        match="All searchers must use a pipeline.",
+        match=r"All searchers must use a pipeline.",
     ):
         merge_pipelines(pipe1, pipe3, search_params=None)  # type: ignore
 
@@ -108,7 +108,7 @@ def test_merger_errors() -> None:
     pipe4 = creator4.to_pipeline(search_params={"kind": "grid"})
     with pytest.raises(
         ValueError,
-        match="must have the same named steps.",
+        match=r"must have the same named steps.",
     ):
         merge_pipelines(pipe1, pipe4, search_params=None)  # type: ignore
 

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/julearn/tests/test_api.py

@@ -1237,7 +1237,7 @@ def test_api_stacking_models() -> None:
         return_estimator="final",
     )
 
-    # The final model should be a stacking model im which the first estimator
+    # The final model should be a stacking model in which the first estimator
     # is a grid search
     assert isinstance(
         final.steps[1][1].estimators[0][1],  # type: ignore

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/julearn/transformers/dataframe/tests/test_set_column_types.py

@@ -59,7 +59,7 @@ def test_SetColumnTypes_input_validation(
 
     """
     with pytest.raises(
-        ValueError, match="Each value of X_types must be a list."
+        ValueError, match=r"Each value of X_types must be a list."
     ):
         SetColumnTypes({"confound": "chicken"}).fit(X_iris)  # type: ignore
 

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/julearn.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: julearn
-Version: 0.3.5.dev56
+Version: 0.3.5.dev67
 Summary: Juelich Machine Learning Library
 Author-email: Fede Raimondo <f.raimondo@fz-juelich.de>, Sami Hamdan <s.hamdan@fz-juelich.de>
 Maintainer-email: Sami Hamdan <s.hamdan@fz-juelich.de>

{julearn-0.3.5.dev56 → julearn-0.3.5.dev67}/julearn.egg-info/SOURCES.txt

@@ -55,6 +55,8 @@ docs/api/utils.rst
 docs/api/viz.rst
 docs/changes/contributors.inc
 docs/changes/newsfragments/.gitignore
+docs/changes/newsfragments/254.bugfix
+docs/changes/newsfragments/254.change
 docs/changes/newsfragments/277.enh
 docs/changes/newsfragments/278.doc
 docs/changes/newsfragments/279.doc

julearn-0.3.5.dev56/julearn/model_selection/stratified_bootstrap.py

@@ -1,145 +0,0 @@
-"""Class for Stratified Bootstrap cross-validation iterator."""
-
-# Authors: Federico Raimondo <f.raimondo@fz-juelich.de>
-#          Sami Hamdan <s.hamdan@fz-juelich.de>
-# License: AGPL
-
-from typing import Optional, Union
-
-import numpy as np
-from numpy.random import RandomState
-from sklearn.model_selection import BaseShuffleSplit
-from sklearn.model_selection._split import (
-    _validate_shuffle_split,  # type: ignore
-)
-
-
-class StratifiedBootstrap(BaseShuffleSplit):
-    """Stratified Bootstrap cross-validation iterator.
-
-    Provides train/test indices using resampling with replacement, respecting
-    the distribution of samples for each class.
-
-    Parameters
-    ----------
-    n_splits : int, default=5
-        Number of re-shuffling & splitting iterations.
-    test_size : float, int, default=0.2
-        If float, should be between 0.0 and 1.0 and represent the proportion
-        of groups to include in the test split (rounded up). If int,
-        represents the absolute number of test groups. If None, the value is
-        set to the complement of the train size.
-        The default will change in version 0.21. It will remain 0.2 only
-        if ``train_size`` is unspecified, otherwise it will complement
-        the specified ``train_size``.
-    train_size : float or int, default=None
-        If float, should be between 0.0 and 1.0 and represent the
-        proportion of the groups to include in the train split. If
-        int, represents the absolute number of train groups. If None,
-        the value is automatically set to the complement of the test size.
-    random_state : int or RandomState instance, default=None
-        Controls the randomness of the training and testing indices produced.
-        Pass an int for reproducible output across multiple function calls.
-
-    """
-
-    def __init__(
-        self,
-        n_splits: int = 5,
-        test_size: float = 0.5,
-        train_size: Optional[float] = None,
-        random_state: Optional[Union[int, RandomState]] = None,
-    ):
-        super().__init__(
-            n_splits=n_splits,
-            test_size=test_size,
-            train_size=train_size,
-            random_state=random_state,
-        )
-
-    def _iter_indices(
-        self,
-        X: np.ndarray,  # noqa: N803
-        y: np.ndarray,
-        groups: Optional[np.ndarray] = None,
-    ):
-        """Generate (train, test) indices.
-
-        Parameters
-        ----------
-        X : array-like of shape (n_samples, n_features)
-            Training vectors, where n_samples is the number of samples
-            and n_features is the number of features.
-        y : array-like of shape (n_samples,)
-            The target variable for supervised learning problems.
-        groups : array-like of shape (n_samples,), default=None
-            Group labels for stratifying the samples used while splitting the
-            dataset into train/test set.
-
-        Yields
-        ------
-        train : ndarray
-            The training set indices for that split.
-        test : ndarray
-            The testing set indices for that split.
-
-        """
-        y_labels = np.unique(y)
-        y_inds = [np.where(y == t_y)[0] for t_y in y_labels]
-        n_samples = [
-            _validate_shuffle_split(
-                len(t_inds),
-                self.test_size,  # type: ignore
-                self.train_size,  # type: ignore
-                default_test_size=self._default_test_size,  # type: ignore
-            )
-            for t_inds in y_inds
-        ]
-        for _ in range(self.n_splits):  # type: ignore
-            train = []
-            test = []
-            for t_inds, (n_train, _) in zip(y_inds, n_samples):
-                bs_inds = np.random.choice(t_inds, len(t_inds), replace=True)
-                train.extend(bs_inds[:n_train])
-                test.extend(bs_inds[n_train:])
-
-            yield train, test
-
-    def split(
-        self,
-        X: np.ndarray,  # noqa: N803
-        y: np.ndarray,
-        groups: Optional[np.ndarray] = None,
-    ):
-        """Generate indices to split data into training and test set.
-
-        Parameters
-        ----------
-        X : array-like of shape (n_samples, n_features)
-            Training data, where n_samples is the number of samples
-            and n_features is the number of features.
-            Note that providing ``y`` is sufficient to generate the splits and
-            hence ``np.zeros(n_samples)`` may be used as a placeholder for
-            ``X`` instead of actual training data.
-        y : array-like of shape (n_samples,) or (n_samples, n_labels)
-            The target variable for supervised learning problems.
-            Stratification is done based on the y labels.
-        groups : array-like of shape (n_samples,), default=None
-            Group labels for stratifying the samples used while splitting the
-            dataset into train/test set.
-
-        Yields
-        ------
-        train : ndarray
-            The training set indices for that split.
-        test : ndarray
-            The testing set indices for that split.
-
-        Notes
-        -----
-        Randomized CV splitters may return different results for each call of
-        split. You can make the results identical by setting `random_state`
-        to an integer.
-
-        """
-        return super().split(X, y, groups)

julearn-0.3.5.dev56/julearn/model_selection/tests/test_stratified_bootstrap.py

@@ -1,46 +0,0 @@
-"""Provides tests for the stratified bootstrap CV generator."""
-
-# Authors: Federico Raimondo <f.raimondo@fz-juelich.de>
-# License: AGPL
-
-import numpy as np
-import pytest
-
-from julearn.model_selection import StratifiedBootstrap
-
-
-@pytest.mark.parametrize(
-    "n_classes, test_size",
-    [
-        (3, 0.2),
-        (2, 0.5),
-        (4, 0.8),
-    ],
-)
-def test_stratified_bootstrap(n_classes: int, test_size: float) -> None:
-    """Test stratified bootstrap CV generator.
-
-    Parameters
-    ----------
-    n_classes : int
-        Number of classes.
-    test_size : float
-        Test size.
-
-    """
-    n_samples = 100
-    X = np.random.rand(n_samples, 2)
-
-    y = np.random.randint(0, n_classes, n_samples)
-
-    cv = StratifiedBootstrap(n_splits=10, test_size=test_size)
-
-    for train, test in cv.split(X, y):
-        y_train = y[train]
-        y_test = y[test]
-        for i in range(n_classes):
-            n_y = (y == i).sum()
-            n_y_train = (y_train == i).sum()
-            n_y_test = (y_test == i).sum()
-            assert abs(n_y_train - (n_y * (1 - test_size))) < 1
-            assert abs(n_y_test - (n_y * test_size)) < 1