julearn 0.3.5.dev117__tar.gz → 0.3.5.dev134__tar.gz

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: julearn
-Version: 0.3.5.dev117
+Version: 0.3.5.dev134
 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>
@@ -27,10 +27,10 @@ Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE.md
 License-File: AUTHORS.rst
-Requires-Dist: numpy<2.0.0,>=1.26.0
-Requires-Dist: pandas<2.4.0,>=2.0.0
+Requires-Dist: numpy<3.0.0,>=1.26.0
+Requires-Dist: pandas<4.0.0,>=2.1.0
 Requires-Dist: statsmodels<0.15,>=0.13
-Requires-Dist: scikit-learn<1.8.0,>=1.5.0
+Requires-Dist: scikit-learn<1.9.0,>=1.5.0
 Requires-Dist: looseversion==1.3.0; python_version >= "3.12"
 Provides-Extra: dev
 Requires-Dist: tox; extra == "dev"

julearn-0.3.5.dev134/docs/changes/newsfragments/298.bugfix

@@ -0,0 +1 @@
+Enable and use metadata routing for hyperparameter tuning estimators by `Fede Raimondo`_

julearn-0.3.5.dev134/docs/changes/newsfragments/299.misc

@@ -0,0 +1 @@
+Update versions of dependencies in pyproject.toml to support pandas 3.X.X, numpy 2.X.X and scikit-learn 1.8.X by `Fede Raimondo`_

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/examples/00_starting/plot_cm_acc_multiclass.py

@@ -89,7 +89,7 @@ print(cm)
 ###############################################################################
 # Now that we have our confusion matrix, let's build another matrix with
 # annotations.
-cm_sum = np.sum(cm, axis=1, keepdims=True)
+cm_sum = np.sum(cm, axis=1)
 cm_perc = cm / cm_sum.astype(float) * 100
 annot = np.empty_like(cm).astype(str)
 nrows, ncols = cm.shape

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/examples/03_complex_models/run_hyperparameter_tuning_bayessearch.py

@@ -19,6 +19,7 @@ References
 
 import numpy as np
 from seaborn import load_dataset
+import sklearn
 
 from julearn import run_cross_validation
 from julearn.utils import configure_logging, logger
@@ -29,6 +30,10 @@ from julearn.pipeline import PipelineCreator
 # Set the logging level to info to see extra information.
 configure_logging(level="INFO")
 
+###############################################################################
+# Disable metadata routing to avoid errors due to BayesSearchCV being used.
+sklearn.set_config(enable_metadata_routing=False)
+
 ###############################################################################
 # Set the random seed to always have the same example.
 np.random.seed(42)

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/examples/99_docs/run_model_inspection_docs.py

@@ -57,6 +57,7 @@ penguins_df = penguins_df.query("species != 'Chinstrap'").dropna()
 penguins_df["species"] = penguins_df["species"].replace(
     {"Adelie": 0, "Gentoo": 1}
 )
+penguins_df["species"] = penguins_df["species"].astype(int)
 features = [x for x in penguins_df.columns if x != "species"]
 
 ##############################################################################

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/_version.py

@@ -28,7 +28,7 @@ version_tuple: VERSION_TUPLE
 commit_id: COMMIT_ID
 __commit_id__: COMMIT_ID
 
-__version__ = version = '0.3.5.dev117'
-__version_tuple__ = version_tuple = (0, 3, 5, 'dev117')
+__version__ = version = '0.3.5.dev134'
+__version_tuple__ = version_tuple = (0, 3, 5, 'dev134')
 
-__commit_id__ = commit_id = 'gb64469b80'
+__commit_id__ = commit_id = 'g92ed90a44'

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/api.py

@@ -4,6 +4,7 @@
 #          Sami Hamdan <s.hamdan@fz-juelich.de>
 # License: AGPL
 
+import inspect
 from typing import Optional, Union
 
 import numpy as np
@@ -26,7 +27,10 @@ from .utils import _compute_cvmdsum, logger, raise_error
 from .utils.typing import CVLike
 
 
-def _validata_api_params(  # noqa: C901
+sklearn.set_config(enable_metadata_routing=True)
+
+
+def _validate_api_params(  # noqa: C901
     X: list[str],  # noqa: N803
     y: str,
     model: Union[str, PipelineCreator, BaseEstimator, list[PipelineCreator]],
@@ -533,7 +537,7 @@ def run_cross_validation(
         return_inspector,
         wrap_score,
         problem_type,
-    ) = _validata_api_params(
+    ) = _validate_api_params(
         X=X,
         y=y,
         model=model,
@@ -561,7 +565,7 @@ def run_cross_validation(
     )
     logger.info(f"Using outer CV scheme {cv_outer}")
 
-    check_consistency(df_y, cv, groups, problem_type)  # type: ignore
+    groups_needed = check_consistency(df_y, cv, groups, problem_type)  # type: ignore
 
     scoring = check_scoring(
         pipeline,  # type: ignore
@@ -570,16 +574,28 @@ def run_cross_validation(
     )
 
     cv_mdsum = _compute_cvmdsum(cv_outer)
-    fit_params = {}
-    if df_groups is not None:
-        if isinstance(pipeline, BaseSearchCV):
-            fit_params["groups"] = df_groups.values
 
+    fit_params = {}
     _sklearn_deprec_fit_params = {}
-    if sklearn.__version__ >= "1.4.0":
-        _sklearn_deprec_fit_params["params"] = fit_params
-    else:
-        _sklearn_deprec_fit_params["fit_params"] = fit_params
+
+    if df_groups is not None:
+        if groups_needed:
+            # If we need groups, we have to pass them to the fit method of
+            # the last step of the pipeline
+            if not isinstance(pipeline, BaseSearchCV):
+                last_step = pipeline.steps[-1][1]
+                argspec = inspect.getfullargspec(last_step.fit)
+                if "groups" in argspec.args:
+                    logger.debug(
+                        "Pipeline's last step uses groups," \
+                        "calling `set_fit_request(groups=True)`")
+                    last_step.set_fit_request(groups=True)
+        fit_params["groups"] = df_groups.values
+
+        if sklearn.__version__ >= "1.4.0":
+            _sklearn_deprec_fit_params["params"] = fit_params
+        else:
+            _sklearn_deprec_fit_params["fit_params"] = fit_params
 
     scores = cross_validate(
         pipeline,
@@ -587,7 +603,6 @@ def run_cross_validation(
         df_y,
         cv=cv_outer,
         scoring=scoring,
-        groups=df_groups,
         return_estimator=cv_return_estimator,
         n_jobs=n_jobs,
         return_train_score=return_train_score,
@@ -776,7 +791,7 @@ def run_fit(
         _,
         _,
         problem_type,
-    ) = _validata_api_params(
+    ) = _validate_api_params(
         X=X,
         y=y,
         model=model,

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/conftest.py

@@ -19,6 +19,14 @@ _filter_keys = {
 }
 
 
+@pytest.fixture(autouse=True)
+def enable_metadata_routing() -> None:
+    """Enable metadata routing in sklearn for all tests."""
+    import sklearn
+
+    sklearn.set_config(enable_metadata_routing=True)
+
+
 def pytest_configure(config: pytest.Config) -> None:
     """Add a new marker to pytest.
 

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/model_selection/final_model_cv.py

@@ -4,16 +4,13 @@
 # License: AGPL
 
 from collections.abc import Generator
-from typing import TYPE_CHECKING, Optional
+from typing import Optional
 
 import numpy as np
+from sklearn.model_selection import BaseCrossValidator
 
 
-if TYPE_CHECKING:
-    from sklearn.model_selection import BaseCrossValidator
-
-
-class _JulearnFinalModelCV:
+class _JulearnFinalModelCV(BaseCrossValidator):
     """Final model cross-validation iterator.
 
     Wraps any CV iterator to provide an extra iteration with the full dataset.
@@ -30,6 +27,20 @@ class _JulearnFinalModelCV:
         if hasattr(cv, "n_repeats"):
             self.n_repeats = cv.n_repeats
 
+    def get_metadata_routing(self) -> dict:
+        """Get metadata routing information from the underlying CV.
+
+        Returns
+        -------
+        dict
+            The metadata routing information.
+
+        """
+        if hasattr(self.cv, "get_metadata_routing"):
+            return self.cv.get_metadata_routing()
+        else:
+            return {}
+
     def split(
         self,
         X: np.ndarray,  # noqa: N803

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/pipeline/pipeline_creator.py

@@ -9,7 +9,9 @@ from dataclasses import dataclass, field
 from typing import Any, Optional, Union
 
 import numpy as np
+import sklearn
 from scipy import stats
+from sklearn.ensemble import AdaBoostClassifier, AdaBoostRegressor
 from sklearn.model_selection import RandomizedSearchCV, check_cv
 from sklearn.pipeline import Pipeline
 
@@ -417,6 +419,16 @@ class PipelineCreator:
             f"{name}__{param}": val for param, val in params_to_tune.items()
         }
 
+        # Disable metadata routing for AdaBoost-based estimators until
+        # scikit-learn implements them.
+
+        if isinstance(step, (AdaBoostClassifier, AdaBoostRegressor)):
+            warn_with_log(
+                "Disabling metadata routing for AdaBoost-based "
+                "estimators until scikit-learn implements them."
+            )
+            sklearn.set_config(enable_metadata_routing=False)
+
         self._steps.append(
             Step(
                 name=name,

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/prepare.py

@@ -365,7 +365,7 @@ def check_consistency(
     cv: CVLike,
     groups: Optional[pd.Series],
     problem_type: str,
-) -> None:
+) -> bool:
     """Check the consistency of the parameters/input.
 
     Parameters
@@ -379,6 +379,12 @@ def check_consistency(
     problem_type : str
         The problem type. Can be "classification" or "regression".
 
+    Returns
+    -------
+    groups_needed : bool
+        True if the groups variable is needed for the CV scheme,
+        False otherwise.
+
     Raises
     ------
     ValueError
@@ -395,6 +401,15 @@ def check_consistency(
     # Check problem type and the target.
     n_classes = np.unique(y.values).shape[0]  # type: ignore
     if problem_type == "classification":
+        is_numeric = isinstance(y.values.dtype, np.dtype) and np.issubdtype(
+            y.values.dtype, np.number
+        )  # type: ignore
+        if not is_numeric:
+            warn_with_log(
+                f"The kind of values in y ({y.values.dtype}) is not "
+                "suitable for a classification. Values should be numeric."
+            )
+
         # If not exactly two classes:
         if n_classes == 1:
             raise_error(
@@ -417,7 +432,9 @@ def check_consistency(
             logger.info("Binary classification problem detected.")
     elif problem_type == "regression":
         # Regression
-        is_numeric = np.issubdtype(y.values.dtype, np.number)  # type: ignore
+        is_numeric = isinstance(y.values.dtype, np.dtype) and np.issubdtype(
+            y.values.dtype, np.number
+        )  # type: ignore
         if not is_numeric:
             warn_with_log(
                 f"The kind of values in y ({y.values.dtype}) is not "
@@ -435,21 +452,30 @@ def check_consistency(
             "The problem type must be either 'classification' or 'regression'."
         )
     # Check groups and CV scheme
+    groups_needed = False
+    valid_group_cv_instances = (
+        GroupKFold,
+        GroupShuffleSplit,
+        LeaveOneGroupOut,
+        LeavePGroupsOut,
+        StratifiedGroupKFold,
+        ContinuousStratifiedGroupKFold,
+        RepeatedContinuousStratifiedGroupKFold,
+    )
     if groups is not None:
-        valid_instances = (
-            GroupKFold,
-            GroupShuffleSplit,
-            LeaveOneGroupOut,
-            LeavePGroupsOut,
-            StratifiedGroupKFold,
-            ContinuousStratifiedGroupKFold,
-            RepeatedContinuousStratifiedGroupKFold,
-        )
-        if not isinstance(cv, valid_instances):
+        if not isinstance(cv, valid_group_cv_instances):
             warn_with_log(
                 "The parameter groups was specified but the CV strategy "
                 "will not consider them."
             )
+        else:
+            groups_needed = True
+    elif isinstance(cv, valid_group_cv_instances):
+        raise_error(
+            "The CV strategy requires groups but the parameter groups was "
+            "not specified."
+        )
+    return groups_needed
 
 
 def _check_x_types(

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/tests/test_api.py

@@ -549,7 +549,6 @@ def test_tune_hyperparam_gridsearch_groups(df_iris: pd.DataFrame) -> None:
             sk_y,  # type: ignore
             cv=cv_outer,
             scoring=[scoring],
-            groups=sk_groups,
             params={"groups": sk_groups},
         )
 

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/tests/test_prepare.py

@@ -701,7 +701,16 @@ def test_check_consistency() -> None:
     with pytest.warns(RuntimeWarning, match="larger than the number"):
         check_consistency(y=y, cv=cv, groups=groups, problem_type=problem_type)
 
-    # Test with wrong problem type
+    # Test with wrong dtype in series
+    y = pd.Series(np.random.randint(0, 2, size=10)).astype(object)
+    problem_type = "classification"
+    groups = None
+    cv = 5
+    with pytest.warns(RuntimeWarning, match="should be numeric"):
+        check_consistency(y=y, cv=cv, groups=groups, problem_type=problem_type)
+
+
+    # Test with wrong dtype in series
     y = pd.Series(["A"] * 10)
     problem_type = "regression"
     groups = None

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/transformers/confound_remover.py

@@ -259,7 +259,7 @@ class ConfoundRemover(JuTransformer):
         if self.threshold is not None:
             # Accounting for correlated rounding errors for very small
             # residuals
-            residuals = residuals.applymap(  # type: ignore
+            residuals = residuals.map(  # type: ignore
                 lambda x: 0 if abs(x) <= self.threshold else x
             )
         return residuals

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn/utils/testing.py

@@ -245,13 +245,17 @@ def do_scoring_test(
     )
 
     np.random.seed(42)
+    if sk_groups is not None:
+        params = {"groups": sk_groups}
+    else:
+        params = {}
     expected = cross_validate(
         sklearn_model,  # type: ignore
         sk_X,
         sk_y,
         cv=sk_cv,
         scoring=scorers,
-        groups=sk_groups,  # type: ignore
+        params=params,  # type: ignore
     )
 
     # Compare the models

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: julearn
-Version: 0.3.5.dev117
+Version: 0.3.5.dev134
 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>
@@ -27,10 +27,10 @@ Requires-Python: >=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE.md
 License-File: AUTHORS.rst
-Requires-Dist: numpy<2.0.0,>=1.26.0
-Requires-Dist: pandas<2.4.0,>=2.0.0
+Requires-Dist: numpy<3.0.0,>=1.26.0
+Requires-Dist: pandas<4.0.0,>=2.1.0
 Requires-Dist: statsmodels<0.15,>=0.13
-Requires-Dist: scikit-learn<1.8.0,>=1.5.0
+Requires-Dist: scikit-learn<1.9.0,>=1.5.0
 Requires-Dist: looseversion==1.3.0; python_version >= "3.12"
 Provides-Extra: dev
 Requires-Dist: tox; extra == "dev"

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn.egg-info/SOURCES.txt

@@ -8,6 +8,7 @@ ignore_words.txt
 pyproject.toml
 setup.py
 tox.ini
+uv.lock
 .github/dependabot.yml
 .github/ISSUE_TEMPLATE/bug_report.yaml
 .github/ISSUE_TEMPLATE/config.yml
@@ -71,6 +72,8 @@ docs/changes/newsfragments/288.bugfix
 docs/changes/newsfragments/291.bugfix
 docs/changes/newsfragments/297.bugfix
 docs/changes/newsfragments/297.misc
+docs/changes/newsfragments/298.bugfix
+docs/changes/newsfragments/299.misc
 docs/images/corrected_ttest.png
 docs/images/final_estimator.png
 docs/images/iris_X.png

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/julearn.egg-info/requires.txt

@@ -1,7 +1,7 @@
-numpy<2.0.0,>=1.26.0
-pandas<2.4.0,>=2.0.0
+numpy<3.0.0,>=1.26.0
+pandas<4.0.0,>=2.1.0
 statsmodels<0.15,>=0.13
-scikit-learn<1.8.0,>=1.5.0
+scikit-learn<1.9.0,>=1.5.0
 
 [:python_version >= "3.12"]
 looseversion==1.3.0

{julearn-0.3.5.dev117 → julearn-0.3.5.dev134}/pyproject.toml

@@ -34,10 +34,10 @@ classifiers = [
     "Programming Language :: Python :: 3.13",
 ]
 dependencies = [
-    "numpy>=1.26.0,<2.0.0",
-    "pandas>=2.0.0,<2.4.0",
+    "numpy>=1.26.0,<3.0.0",
+    "pandas>=2.1.0,<4.0.0",
     "statsmodels>=0.13,<0.15",
-    "scikit-learn>=1.5.0,<1.8.0",
+    "scikit-learn>=1.5.0,<1.9.0",
     "looseversion==1.3.0; python_version>='3.12'",
 ]
 dynamic = ["version"]