scikit-learn · ogrisel · Jun 24, 2020 · Jun 11, 2020 · Jun 11, 2020 · Jun 11, 2020
diff --git a/doc/whats_new/v0.24.rst b/doc/whats_new/v0.24.rst
@@ -117,6 +117,11 @@ Changelog
   :func:`metrics.median_absolute_error`. :pr:`17225` by
   :user:`Lucy Liu <lucyleeow>`.
 
+- |Enhancement| Add `pos_label` parameter in
+  :func:`metrics.plot_precision_recall_curve` in order to specify the positive
+  class to be used when computing the precision and recall statistics.
+  :pr:`17569` by :user:`Guillaume Lemaitre <glemaitre>`.
+
 - |Enhancement| Add `pos_label` parameter to :func:`roc_auc_score`.
   :pr:`17594` by :user:`Guillaume Lemaitre <glemaitre>`.
 

diff --git a/sklearn/metrics/_plot/precision_recall_curve.py b/sklearn/metrics/_plot/precision_recall_curve.py
@@ -1,3 +1,5 @@
+import numpy as np
+
 from .base import _check_classifier_response_method
 
 from .. import average_precision_score
@@ -30,6 +32,12 @@ class PrecisionRecallDisplay:
     estimator_name : str, default=None
         Name of estimator. If None, then the estimator name is not shown.
 
+    pos_label : str or int, default=None
+        The class considered as the positive class. If None, the class will not
+        be shown in the legend.
+
+        .. versionadded:: 0.24
+
     Attributes
     ----------
     line_ : matplotlib Artist
@@ -60,11 +68,12 @@ class PrecisionRecallDisplay:
     >>> disp.plot() # doctest: +SKIP
     """
     def __init__(self, precision, recall, *,
-                 average_precision=None, estimator_name=None):
+                 average_precision=None, estimator_name=None, pos_label=None):
         self.precision = precision
         self.recall = recall
         self.average_precision = average_precision
         self.estimator_name = estimator_name
+        self.pos_label = pos_label
 
     @_deprecate_positional_args
     def plot(self, ax=None, *, name=None, **kwargs):
@@ -110,7 +119,11 @@ def plot(self, ax=None, *, name=None, **kwargs):
         line_kwargs.update(**kwargs)
 
         self.line_, = ax.plot(self.recall, self.precision, **line_kwargs)
-        ax.set(xlabel="Recall", ylabel="Precision")
+        info_pos_label = (f" (Positive label: {self.pos_label})"
+                          if self.pos_label is not None else "")
+        xlabel = "Recall" + info_pos_label
+        ylabel = "Precision" + info_pos_label
+        ax.set(xlabel=xlabel, ylabel=ylabel)
 
         if "label" in line_kwargs:
             ax.legend(loc='lower left')
@@ -123,7 +136,7 @@ def plot(self, ax=None, *, name=None, **kwargs):
 @_deprecate_positional_args
 def plot_precision_recall_curve(estimator, X, y, *,
                                 sample_weight=None, response_method="auto",
-                                name=None, ax=None, **kwargs):
+                                name=None, ax=None, pos_label=None, **kwargs):
     """Plot Precision Recall Curve for binary classifiers.
 
     Extra keyword arguments will be passed to matplotlib's `plot`.
@@ -159,6 +172,13 @@ def plot_precision_recall_curve(estimator, X, y, *,
     ax : matplotlib axes, default=None
         Axes object to plot on. If `None`, a new figure and axes is created.
 
+    pos_label : str or int, default=None
+        The class considered as the positive class when computing the precision
+        and recall metrics. By default, `estimators.classes_[1]` is considered
+        as the positive class.
+
+        .. versionadded:: 0.24
+
     **kwargs : dict
         Keyword arguments to be passed to matplotlib's `plot`.
 
@@ -180,16 +200,30 @@ def plot_precision_recall_curve(estimator, X, y, *,
         raise ValueError(classification_error)
 
     prediction_method = _check_classifier_response_method(estimator,
-                                                         response_method)
+                                                          response_method)
     y_pred = prediction_method(X)
 
-    if y_pred.ndim != 1:
+    if pos_label is not None and pos_label not in estimator.classes_:
+        raise ValueError(
+            f"The class provided by 'pos_label' is unknown. Got "
+            f"{pos_label} instead of one of {estimator.classes_}"
+        )
+
+    if y_pred.ndim != 1:  # `predict_proba`
         if y_pred.shape[1] != 2:
             raise ValueError(classification_error)
-        else:
+        if pos_label is None:
+            pos_label = estimator.classes_[1]
             y_pred = y_pred[:, 1]
+        else:
+            class_idx = np.flatnonzero(estimator.classes_ == pos_label)
+            y_pred = y_pred[:, class_idx]
+    else:  # `decision_function`
+        if pos_label is None:
+            pos_label = estimator.classes_[1]
+        elif pos_label == estimator.classes_[0]:
+            y_pred *= -1
 
-    pos_label = estimator.classes_[1]
     precision, recall, _ = precision_recall_curve(y, y_pred,
                                                   pos_label=pos_label,
                                                   sample_weight=sample_weight)
@@ -199,6 +233,7 @@ def plot_precision_recall_curve(estimator, X, y, *,
     name = name if name is not None else estimator.__class__.__name__
     viz = PrecisionRecallDisplay(
         precision=precision, recall=recall,
-        average_precision=average_precision, estimator_name=name
+        average_precision=average_precision, estimator_name=name,
+        pos_label=pos_label,
     )
     return viz.plot(ax=ax, name=name, **kwargs)
diff --git a/sklearn/metrics/_plot/tests/test_plot_precision_recall.py b/sklearn/metrics/_plot/tests/test_plot_precision_recall.py
@@ -11,9 +11,11 @@
 from sklearn.datasets import load_breast_cancer
 from sklearn.tree import DecisionTreeClassifier, DecisionTreeRegressor
 from sklearn.linear_model import LogisticRegression
+from sklearn.model_selection import train_test_split
 from sklearn.exceptions import NotFittedError
 from sklearn.pipeline import make_pipeline
 from sklearn.preprocessing import StandardScaler
+from sklearn.utils import shuffle
 from sklearn.compose import make_column_transformer
 
 
@@ -114,8 +116,8 @@ def test_plot_precision_recall(pyplot, response_method, with_sample_weight):
 
     expected_label = "LogisticRegression (AP = {:0.2f})".format(avg_prec)
     assert disp.line_.get_label() == expected_label
-    assert disp.ax_.get_xlabel() == "Recall"
-    assert disp.ax_.get_ylabel() == "Precision"
+    assert disp.ax_.get_xlabel() == "Recall (Positive label: 1)"
+    assert disp.ax_.get_ylabel() == "Precision (Positive label: 1)"
 
     # draw again with another label
     disp.plot(name="MySpecialEstimator")
@@ -190,3 +192,50 @@ def test_default_labels(pyplot, average_precision, estimator_name,
                                   estimator_name=estimator_name)
     disp.plot()
     assert disp.line_.get_label() == expected_label
+
+
+@pytest.mark.parametrize(
+    "response_method", ["predict_proba", "decision_function"]
+)
+def test_plot_precision_recall_pos_label(pyplot, response_method):
+    # check that we can provide the positive label and display the proper
+    # statistics
+    X, y = load_breast_cancer(return_X_y=True)
+    # create an highly imbalanced version of the breast cancer dataset
+    idx_positive = np.flatnonzero(y == 1)
+    idx_negative = np.flatnonzero(y == 0)
+    idx_selected = np.hstack([idx_negative, idx_positive[:25]])
+    X, y = X[idx_selected], y[idx_selected]
+    X, y = shuffle(X, y, random_state=42)
+    # only use 2 features to make the problem even harder
+    X = X[:, :2]
+    y = np.array(
+        ["cancer" if c == 1 else "not cancer" for c in y], dtype=object
+    )
+    X_train, X_test, y_train, y_test = train_test_split(
+        X, y, stratify=y, random_state=0,
+    )
+
+    classifier = LogisticRegression()
+    classifier.fit(X_train, y_train)
+
+    # sanity check to be sure the positive class is classes_[0] and that we
+    # are betrayed by the class imbalance
+    assert classifier.classes_.tolist() == ["cancer", "not cancer"]
+
+    disp = plot_precision_recall_curve(
+        classifier, X_test, y_test, pos_label="cancer",
+        response_method=response_method
+    )
+    # we should obtain the statistics of the "cancer" class
+    avg_prec_limit = 0.65
+    assert disp.average_precision < avg_prec_limit
+    assert -np.trapz(disp.precision, disp.recall) < avg_prec_limit
+
+    # otherwise we should obtain the statistics of the "not cancer" class
+    disp = plot_precision_recall_curve(
+        classifier, X_test, y_test, response_method=response_method,
+    )
+    avg_prec_limit = 0.95
+    assert disp.average_precision > avg_prec_limit
+    assert -np.trapz(disp.precision, disp.recall) > avg_prec_limit