ENH Add pos_label parameter to TargetEncoder #27342

[itonito]

Reference Issues/PRs

Fixes #27342

What does this implement/fix? Explain your changes.

Added pos_label and tests

Any other comments?

[github-actions]

❌ Linting issues

This PR is introducing linting issues. Here's a summary of the issues. Note that you can avoid having linting issues by enabling pre-commit hooks. Instructions to enable them can be found here.

You can see the details of the linting issues under the lint job here

---

ruff format

ruff detected issues. Please run ruff format locally and push the changes. Here you can see the detected issues. Note that the installed ruff version is ruff=0.11.7.

--- sklearn/preprocessing/tests/test_target_encoder.py
+++ sklearn/preprocessing/tests/test_target_encoder.py
@@ -722,28 +722,20 @@
     y_binary = np.array([0, 1, 0, 1, 0, 1])
 
     # Default pos_label=1 should give same result as LabelBinarizer default
-    enc_default = TargetEncoder(
-        target_type="binary", cv=2, random_state=0
-    )
+    enc_default = TargetEncoder(target_type="binary", cv=2, random_state=0)
     enc_default.fit(X, y_binary)
 
     # Custom pos_label=0 should invert the encoding
-    enc_custom = TargetEncoder(
-        target_type="binary", pos_label=0, cv=2, random_state=0
-    )
+    enc_custom = TargetEncoder(target_type="binary", pos_label=0, cv=2, random_state=0)
     enc_custom.fit(X, y_binary)
 
     # The encodings should be different due to different pos_label
-    assert not np.allclose(
-        enc_default.encodings_[0], enc_custom.encodings_[0]
-    )
+    assert not np.allclose(enc_default.encodings_[0], enc_custom.encodings_[0])
 
     # Test multiclass classification with different pos_label values
     y_multiclass = np.array([0, 1, 2, 0, 1, 2])
 
-    enc_multi_default = TargetEncoder(
-        target_type="multiclass", cv=2, random_state=0
-    )
+    enc_multi_default = TargetEncoder(target_type="multiclass", cv=2, random_state=0)
     enc_multi_default.fit(X, y_multiclass)
 
     enc_multi_custom = TargetEncoder(
@@ -768,31 +760,22 @@
     y = np.array([0, 1, 0, 1, 0, 1])
 
     # Test with pos_label=1 (default)
-    enc_te = TargetEncoder(
-        target_type="binary", cv=2, random_state=0
-    )
+    enc_te = TargetEncoder(target_type="binary", cv=2, random_state=0)
     enc_te.fit(X, y)
 
     # Test with pos_label=0
-    enc_te_0 = TargetEncoder(
-        target_type="binary", pos_label=0, cv=2, random_state=0
-    )
+    enc_te_0 = TargetEncoder(target_type="binary", pos_label=0, cv=2, random_state=0)
     enc_te_0.fit(X, y)
 
     # The encodings should be different but the feature names should be consistent
-    assert not np.allclose(
-        enc_te.encodings_[0], enc_te_0.encodings_[0]
-    )
+    assert not np.allclose(enc_te.encodings_[0], enc_te_0.encodings_[0])
 
     # Both should have the same classes
     assert_array_equal(enc_te.classes_, enc_te_0.classes_)
 
     # Test transform output shapes are the same
     X_test = np.array([["a", "b", "c"]], dtype=object).T
-    assert (
-        enc_te.transform(X_test).shape
-        == enc_te_0.transform(X_test).shape
-    )
+    assert enc_te.transform(X_test).shape == enc_te_0.transform(X_test).shape
 
 
 def test_pos_label_parameter_edge_cases():
@@ -806,9 +789,7 @@
     y = np.array([0, 1, 0, 1])
 
     # Test with pos_label that doesn't exist in y (should still work)
-    enc = TargetEncoder(
-        target_type="binary", pos_label=99, cv=2, random_state=0
-    )
+    enc = TargetEncoder(target_type="binary", pos_label=99, cv=2, random_state=0)
     enc.fit(X, y)
 
     # Should still encode correctly
@@ -817,9 +798,7 @@
 
     # Test with pos_label that is a string (should work if y contains strings)
     y_str = np.array(["neg", "pos", "neg", "pos"])
-    enc_str = TargetEncoder(
-        target_type="binary", pos_label="pos", cv=2, random_state=0
-    )
+    enc_str = TargetEncoder(target_type="binary", pos_label="pos", cv=2, random_state=0)
     enc_str.fit(X, y_str)
 
     # Should encode correctly

1 file would be reformatted, 925 files already formatted

_{Generated for commit: ad4ee4e. Link to the linter CI: here}