MTN: move v1 utils in mapie/utils.py, make all utils private except train_conformalize_test_split

Author: Valentin-Laurent

doc/api.rst

@@ -98,7 +98,7 @@ Utils (from mapie_v1)
    :toctree: generated/
    :template: function.rst
 
-   mapie_v1.utils.train_conformalize_test_split
+   mapie.utils.train_conformalize_test_split
 
 Conformity Scores (Regression)
 ==============================

doc/split_cross_conformal.rst

@@ -18,7 +18,7 @@ In all cases, the training/conformalization process can be broken down as follow
 ==============================
 
 - Compute conformity scores ("conformalization") on a conformity set not seen by the model during training.
-  (Use :func:`~mapie_v1.utils.train_conformalize_test_split` to obtain the different sets.)
+  (Use :func:`~mapie.utils.train_conformalize_test_split` to obtain the different sets.)
 
 **MAPIE** then uses the conformity scores to estimate sets associated with the desired coverage on new data with strong theoretical guarantees.
 

examples/mondrian/1-quickstart/plot_main-tutorial-mondrian-regression.py

@@ -45,7 +45,7 @@
 from sklearn.ensemble import RandomForestRegressor
 
 from mapie.metrics.regression import regression_coverage_score
-from mapie_v1.utils import train_conformalize_test_split
+from mapie.utils import train_conformalize_test_split
 from mapie.regression import SplitConformalRegressor
 
 os.environ["TF_CPP_MIN_LOG_LEVEL"] = "3"

examples/regression/1-quickstart/plot_prefit.py

@@ -29,7 +29,7 @@
 from numpy.typing import NDArray
 from mapie.metrics.regression import regression_coverage_score
 from mapie.regression import SplitConformalRegressor, ConformalizedQuantileRegressor
-from mapie_v1.utils import train_conformalize_test_split
+from mapie.utils import train_conformalize_test_split
 
 warnings.filterwarnings("ignore")
 

examples/regression/1-quickstart/plot_toy_model.py

@@ -11,7 +11,7 @@
 
 from mapie.metrics.regression import regression_coverage_score
 from mapie.regression import SplitConformalRegressor
-from mapie_v1.utils import train_conformalize_test_split
+from mapie.utils import train_conformalize_test_split
 
 RANDOM_STATE = 42
 

examples/regression/2-advanced-analysis/plot_ResidualNormalisedScore_tutorial.py

@@ -26,7 +26,7 @@
     regression_ssc_score,
 )
 from mapie.regression import SplitConformalRegressor
-from mapie_v1.utils import train_conformalize_test_split
+from mapie.utils import train_conformalize_test_split
 
 warnings.filterwarnings("ignore")
 

examples/regression/2-advanced-analysis/plot_conformal_predictive_distribution.py

@@ -27,7 +27,7 @@
 from mapie.conformity_scores import (AbsoluteConformityScore,
                                      ResidualNormalisedScore)
 from mapie.regression import SplitConformalRegressor
-from mapie_v1.utils import train_conformalize_test_split
+from mapie.utils import train_conformalize_test_split
 
 warnings.filterwarnings('ignore')
 

examples/regression/2-advanced-analysis/plot_cqr_symmetry_difference.py

@@ -14,7 +14,7 @@
 
 from mapie.metrics.regression import regression_coverage_score
 from mapie.regression import ConformalizedQuantileRegressor
-from mapie_v1.utils import train_conformalize_test_split
+from mapie.utils import train_conformalize_test_split
 
 RANDOM_STATE = 1
 

mapie/__init__.py

@@ -1,9 +1,10 @@
-from . import classification, metrics, regression
+from . import classification, metrics, regression, utils
 from ._version import __version__
 
 __all__ = [
     "regression",
     "classification",
     "metrics",
+    "utils",
     "__version__"
 ]

mapie/calibration.py

@@ -13,9 +13,9 @@
                                       indexable)
 
 from numpy.typing import ArrayLike, NDArray
-from .utils import (check_estimator_classification,
-                    check_estimator_fit_predict, check_n_features_in,
-                    check_null_weight, fit_estimator, get_calib_set)
+from .utils import (_check_estimator_classification,
+                    _check_estimator_fit_predict, _check_n_features_in,
+                    _check_null_weight, _fit_estimator, _get_calib_set)
 
 
 class MapieCalibrator(BaseEstimator, ClassifierMixin):
@@ -202,7 +202,7 @@ def _check_calibrator(
                     "Please provide a string in: "
                     + (", ").join(self.named_calibrators.keys()) + "."
                 )
-        check_estimator_fit_predict(calibrator)
+        _check_estimator_fit_predict(calibrator)
         return calibrator
 
     def _get_labels(
@@ -313,14 +313,14 @@ def _fit_calibrator(
             sample_weight_ = sample_weight[given_label_indices]
             (
                 sample_weight_, top_class_prob_, y_calib_
-            ) = check_null_weight(
+            ) = _check_null_weight(
                 sample_weight_,
                 top_class_prob_,
                 y_calib_
             )
         else:
             sample_weight_ = sample_weight
-        calibrator_ = fit_estimator(
+        calibrator_ = _fit_estimator(
             calibrator_, top_class_prob_, y_calib_, sample_weight_
         )
         return calibrator_
@@ -479,10 +479,10 @@ def fit(
         X, y = indexable(X, y)
         y = _check_y(y)
         self._check_type_of_target(y)
-        estimator = check_estimator_classification(X, y, cv, self.estimator)
+        estimator = _check_estimator_classification(X, y, cv, self.estimator)
         calibrator = self._check_calibrator(self.calibrator)
-        sample_weight, X, y = check_null_weight(sample_weight, X, y)
-        self.n_features_in_ = check_n_features_in(X, cv, estimator)
+        sample_weight, X, y = _check_null_weight(sample_weight, X, y)
+        self.n_features_in_ = _check_n_features_in(X, cv, estimator)
         random_state = check_random_state(random_state)
 
         if cv == "prefit":
@@ -493,7 +493,7 @@ def fit(
                 X, y, sample_weight, calibrator
             )
         if cv == "split":
-            results = get_calib_set(
+            results = _get_calib_set(
                 X,
                 y,
                 sample_weight=sample_weight,
@@ -505,12 +505,12 @@ def fit(
             X_train, y_train, X_calib, y_calib, sw_train, sw_calib = results
             X_train, y_train = indexable(X_train, y_train)
             y_train = _check_y(y_train)
-            sw_train, X_train, y_train = check_null_weight(
+            sw_train, X_train, y_train = _check_null_weight(
                 sw_train,
                 X_train,
                 y_train
             )
-            estimator = fit_estimator(
+            estimator = _fit_estimator(
                 clone(estimator), X_train, y_train, sw_train, **fit_params,
             )
             self.single_estimator_ = estimator