Merge pull request #427 from scikit-learn-contrib/426-move-some-metho…

…ds-of-mapieclassifier-to-a-utils-file

ENH: move get_true_label_position to utils in classification

mapie/classification.py

@@ -23,6 +23,11 @@
                     compute_quantiles, fit_estimator, fix_number_of_classes)
 
 
+from mapie.conformity_scores.utils_classification_conformity_scores import (
+    get_true_label_position,
+)
+
+
 class MapieClassifier(BaseEstimator, ClassifierMixin):
     """
     Prediction sets for classification.
@@ -737,39 +742,6 @@ def _regularize_conformity_score(
         )
         return conf_score
 
-    def _get_true_label_position(
-        self,
-        y_pred_proba: NDArray,
-        y: NDArray
-    ) -> NDArray:
-        """
-        Return the sorted position of the true label in the
-        prediction
-
-        Parameters
-        ----------
-        y_pred_proba: NDArray of shape (n_samples, n_calsses)
-            Model prediction.
-
-        y: NDArray of shape (n_samples)
-            Labels.
-
-        Returns
-        -------
-        NDArray of shape (n_samples, 1)
-            Position of the true label in the prediction.
-        """
-        index = np.argsort(
-                np.fliplr(np.argsort(y_pred_proba, axis=1))
-            )
-        position = np.take_along_axis(
-            index,
-            y.reshape(-1, 1),
-            axis=1
-        )
-
-        return position
-
     def _get_last_included_proba(
         self,
         y_pred_proba: NDArray,
@@ -1217,7 +1189,7 @@ def fit(
             self.y_pred_proba_raps = self.single_estimator_.predict_proba(
                 self.X_raps
             )
-            self.position_raps = self._get_true_label_position(
+            self.position_raps = get_true_label_position(
                 self.y_pred_proba_raps,
                 self.y_raps
             )
@@ -1249,7 +1221,7 @@ def fit(
             # Here we reorder the labels by decreasing probability
             # and get the position of each label from decreasing
             # probability
-            self.conformity_scores_ = self._get_true_label_position(
+            self.conformity_scores_ = get_true_label_position(
                 y_pred_proba,
                 y_enc
             )

mapie/conformity_scores/utils_classification_conformity_scores.py

@@ -0,0 +1,26 @@
+import numpy as np
+from mapie._typing import NDArray
+
+
+def get_true_label_position(y_pred_proba: NDArray, y: NDArray) -> NDArray:
+    """
+    Return the sorted position of the true label in the
+    prediction
+
+    Parameters
+    ----------
+    y_pred_proba: NDArray of shape (n_samples, n_classes)
+        Model prediction.
+
+    y: NDArray of shape (n_samples)
+        Labels.
+
+    Returns
+    -------
+    NDArray of shape (n_samples, 1)
+        Position of the true label in the prediction.
+    """
+    index = np.argsort(np.fliplr(np.argsort(y_pred_proba, axis=1)))
+    position = np.take_along_axis(index, y.reshape(-1, 1), axis=1)
+
+    return position

mapie/tests/test_classification.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from copy import deepcopy
-from typing import Any, Dict, Iterable, List, Optional, Union, cast
+from typing import Any, Dict, Iterable, Optional, Union, cast
 
 import numpy as np
 import pandas as pd
@@ -57,25 +57,6 @@
     )
 ]
 
-Y_TRUE_PROBA_PLACE = [
-    [
-        np.array([2, 0]),
-        np.array([
-            [.1, .3, .6],
-            [.2, .7, .1]
-        ]),
-        np.array([[0], [1]])
-    ],
-    [
-        np.array([1, 0]),
-        np.array([
-            [.7, .12, .18],
-            [.5, .24, .26]
-        ]),
-        np.array([[2], [0]])
-    ]
-]
-
 Params = TypedDict(
     "Params",
     {
@@ -1854,23 +1835,6 @@ def test_get_last_included_proba_shape(k_lambda, strategy):
     assert y_p_p_i_l.shape == (len(X), 1, len(thresholds))
 
 
-@pytest.mark.parametrize("y_true_proba_place", Y_TRUE_PROBA_PLACE)
-def test_get_true_label_position(
-    y_true_proba_place: List[NDArray]
-) -> None:
-    """
-    Check that the returned true label position the good.
-    """
-    y_true = y_true_proba_place[0]
-    y_pred_proba = y_true_proba_place[1]
-    place = y_true_proba_place[2]
-
-    mapie = MapieClassifier(random_state=random_state)
-    found_place = mapie._get_true_label_position(y_pred_proba, y_true)
-
-    assert (found_place == place).all()
-
-
 @pytest.mark.parametrize("cv", [5, None])
 def test_error_raps_cv_not_prefit(cv: Union[int, None]) -> None:
     """

mapie/tests/test_utils_classification_conformity_scores.py

@@ -0,0 +1,54 @@
+from typing import List
+
+import numpy as np
+import pytest
+
+from mapie.conformity_scores.utils_classification_conformity_scores import (
+    get_true_label_position,
+)
+from mapie._typing import NDArray
+
+Y_TRUE_PROBA_PLACE = [
+    [
+        np.array([2, 0]),
+        np.array([
+            [.1, .3, .6],
+            [.2, .7, .1]
+        ]),
+        np.array([[0], [1]])
+    ],
+    [
+        np.array([1, 0]),
+        np.array([
+            [.7, .12, .18],
+            [.5, .24, .26]
+        ]),
+        np.array([[2], [0]])
+    ]
+]
+
+
+def test_shape_get_true_label_position() -> None:
+    """
+    Check the shape returned by the function
+    """
+    y_pred_proba = np.random.rand(5, 3)
+    y = np.random.randint(0, 3, size=(5, 1))
+    position = get_true_label_position(y_pred_proba, y)
+    assert position.shape == y.shape
+
+
+@pytest.mark.parametrize("y_true_proba_place", Y_TRUE_PROBA_PLACE)
+def test_get_true_label_position(
+    y_true_proba_place: List[NDArray]
+) -> None:
+    """
+    Check that the returned true label position the good.
+    """
+    y_true = y_true_proba_place[0]
+    y_pred_proba = y_true_proba_place[1]
+    place = y_true_proba_place[2]
+
+    found_place = get_true_label_position(y_pred_proba, y_true)
+
+    assert (found_place == place).all()