Test.

Author: trivialfis

python-package/xgboost/sklearn.py

@@ -1,5 +1,4 @@
-# coding: utf-8
-# pylint: disable=too-many-arguments, too-many-locals, invalid-name, fixme, R0912, C0302
+# pylint: disable=too-many-arguments, too-many-locals, invalid-name, fixme, too-many-lines
 """Scikit-Learn Wrapper interface for XGBoost."""
 import copy
 import warnings
@@ -544,8 +543,8 @@ def get_xgb_params(self) -> Dict[str, Any]:
         params = self.get_params()
         # Parameters that should not go into native learner.
         wrapper_specific = {
-            'importance_type', 'kwargs', 'missing', 'n_estimators',
-            "enable_categorical", "use_label_encoder",
+            'importance_type', 'kwargs', 'missing', 'n_estimators', 'use_label_encoder',
+            "enable_categorical"
         }
         filtered = {}
         for k, v in params.items():
@@ -1148,33 +1147,31 @@ def fit(
     ) -> "XGBClassifier":
         # pylint: disable = attribute-defined-outside-init,too-many-statements
         evals_result: TrainingCallback.EvalsLog = {}
+
         if _is_cudf_df(y) or _is_cudf_ser(y):
             import cupy as cp  # pylint: disable=E0401
 
             self.classes_ = cp.unique(y.values)
             self.n_classes_ = len(self.classes_)
             expected_classes = cp.arange(self.n_classes_)
-            if (
-                self.classes_.shape != expected_classes.shape
-                or not (self.classes_ == expected_classes).all()
-            ):
-                raise ValueError("Invalid classes")
         elif _is_cupy_array(y):
             import cupy as cp  # pylint: disable=E0401
 
             self.classes_ = cp.unique(y)
             self.n_classes_ = len(self.classes_)
             expected_classes = cp.arange(self.n_classes_)
-            if (
-                self.classes_.shape != expected_classes.shape
-                or not (self.classes_ == expected_classes).all()
-            ):
-                raise ValueError("Invalid classes")
         else:
             self.classes_ = np.unique(np.asarray(y))
             self.n_classes_ = len(self.classes_)
-            if not np.array_equal(self.classes_, np.arange(self.n_classes_)):
-                raise ValueError("Invalid classes")
+            expected_classes = np.arange(self.n_classes_)
+        if (
+            self.classes_.shape != expected_classes.shape
+            or not (self.classes_ == expected_classes).all()
+        ):
+            raise ValueError(
+                f"Invalid classes inferred from unique values of `y`.  "
+                f"Expected: {expected_classes}, got {self.classes_}"
+            )
 
         params = self.get_xgb_params()
 

tests/python-gpu/test_from_cudf.py

@@ -208,7 +208,7 @@ def test_cudf_training_with_sklearn():
     y_cudf_series = ss(data=y.iloc[:, 0])
 
     for y_obj in [y_cudf, y_cudf_series]:
-        clf = xgb.XGBClassifier(gpu_id=0, tree_method='gpu_hist', use_label_encoder=False)
+        clf = xgb.XGBClassifier(gpu_id=0, tree_method='gpu_hist')
         clf.fit(X_cudf, y_obj, sample_weight=cudf_weights, base_margin=cudf_base_margin,
                 eval_set=[(X_cudf, y_obj)])
         pred = clf.predict(X_cudf)

tests/python-gpu/test_from_cupy.py

@@ -122,7 +122,7 @@ def test_cupy_training_with_sklearn():
     base_margin = np.random.random(50)
     cupy_base_margin = cp.array(base_margin)
 
-    clf = xgb.XGBClassifier(gpu_id=0, tree_method="gpu_hist", use_label_encoder=False)
+    clf = xgb.XGBClassifier(gpu_id=0, tree_method="gpu_hist")
     clf.fit(
         X,
         y,

tests/python-gpu/test_gpu_basic_models.py

@@ -7,23 +7,20 @@
 # Don't import the test class, otherwise they will run twice.
 import test_callback as test_cb  # noqa
 import test_basic_models as test_bm
+import testing as tm
 rng = np.random.RandomState(1994)
 
 
 class TestGPUBasicModels:
     cpu_test_cb = test_cb.TestCallbacks()
     cpu_test_bm = test_bm.TestModels()
 
-    def run_cls(self, X, y, deterministic):
-        cls = xgb.XGBClassifier(tree_method='gpu_hist',
-                                deterministic_histogram=deterministic,
-                                single_precision_histogram=True)
+    def run_cls(self, X, y):
+        cls = xgb.XGBClassifier(tree_method='gpu_hist', single_precision_histogram=True)
         cls.fit(X, y)
         cls.get_booster().save_model('test_deterministic_gpu_hist-0.json')
 
-        cls = xgb.XGBClassifier(tree_method='gpu_hist',
-                                deterministic_histogram=deterministic,
-                                single_precision_histogram=True)
+        cls = xgb.XGBClassifier(tree_method='gpu_hist', single_precision_histogram=True)
         cls.fit(X, y)
         cls.get_booster().save_model('test_deterministic_gpu_hist-1.json')
 
@@ -49,19 +46,22 @@ def test_deterministic_gpu_hist(self):
         kClasses = 4
         # Create large values to force rounding.
         X = np.random.randn(kRows, kCols) * 1e4
-        y = np.random.randint(0, kClasses, size=kRows) * 1e4
+        y = np.random.randint(0, kClasses, size=kRows)
 
-        model_0, model_1 = self.run_cls(X, y, True)
+        model_0, model_1 = self.run_cls(X, y)
         assert model_0 == model_1
 
+    @pytest.mark.skipif(**tm.no_sklearn())
     def test_invalid_gpu_id(self):
-        X = np.random.randn(10, 5) * 1e4
-        y = np.random.randint(0, 2, size=10) * 1e4
+        from sklearn.datasets import load_digits
+        X, y = load_digits(return_X_y=True)
         # should pass with invalid gpu id
         cls1 = xgb.XGBClassifier(tree_method='gpu_hist', gpu_id=9999)
         cls1.fit(X, y)
         # should throw error with fail_on_invalid_gpu_id enabled
-        cls2 = xgb.XGBClassifier(tree_method='gpu_hist', gpu_id=9999, fail_on_invalid_gpu_id=True)
+        cls2 = xgb.XGBClassifier(
+            tree_method='gpu_hist', gpu_id=9999, fail_on_invalid_gpu_id=True
+        )
         try:
             cls2.fit(X, y)
             assert False, "Should have failed with with fail_on_invalid_gpu_id enabled"

tests/python-gpu/test_gpu_pickling.py

@@ -146,6 +146,7 @@ def test_pickled_predictor(self):
 
         os.remove(model_path)
 
+    @pytest.mark.skipif(**tm.no_sklearn())
     def test_predict_sklearn_pickle(self):
         from sklearn.datasets import load_digits
         x, y = load_digits(return_X_y=True)

tests/python-gpu/test_gpu_with_sklearn.py

@@ -56,7 +56,6 @@ def test_categorical():
     X, y = load_svmlight_file(os.path.join(data_dir, "agaricus.txt.train"))
     clf = xgb.XGBClassifier(
         tree_method="gpu_hist",
-        use_label_encoder=False,
         enable_categorical=True,
         n_estimators=10,
     )
@@ -98,3 +97,36 @@ def check_predt(X, y):
 
     X = cudf.DataFrame(X)
     check_predt(X, y)
+
+
+@pytest.mark.skipif(**tm.no_cupy())
+@pytest.mark.skipif(**tm.no_cudf())
+def test_classififer():
+    from sklearn.datasets import load_digits
+    import cupy as cp
+    import cudf
+
+    X, y = load_digits(return_X_y=True)
+    y *= 10
+
+    clf = xgb.XGBClassifier(tree_method="gpu_hist", n_estimators=1)
+
+    # numpy
+    with pytest.raises(ValueError, match=r"Invalid classes.*"):
+        clf.fit(X, y)
+
+    # cupy
+    X, y = cp.array(X), cp.array(y)
+    with pytest.raises(ValueError, match=r"Invalid classes.*"):
+        clf.fit(X, y)
+
+    # cudf
+    X, y = cudf.DataFrame(X), cudf.DataFrame(y)
+    with pytest.raises(ValueError, match=r"Invalid classes.*"):
+        clf.fit(X, y)
+
+    # pandas
+    X, y = load_digits(return_X_y=True, as_frame=True)
+    y *= 10
+    with pytest.raises(ValueError, match=r"Invalid classes.*"):
+        clf.fit(X, y)

tests/python/test_with_sklearn.py

@@ -283,7 +283,6 @@ def test_feature_importances_gain():
         random_state=0, tree_method="exact",
         learning_rate=0.1,
         importance_type="gain",
-        use_label_encoder=False,
     ).fit(X, y)
 
     exp = np.array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
@@ -306,7 +305,6 @@ def test_feature_importances_gain():
         tree_method="exact",
         learning_rate=0.1,
         importance_type="gain",
-        use_label_encoder=False,
     ).fit(X, y)
     np.testing.assert_almost_equal(xgb_model.feature_importances_, exp)
 
@@ -315,14 +313,11 @@ def test_feature_importances_gain():
         tree_method="exact",
         learning_rate=0.1,
         importance_type="gain",
-        use_label_encoder=False,
     ).fit(X, y)
     np.testing.assert_almost_equal(xgb_model.feature_importances_, exp)
 
     # no split can be found
-    cls = xgb.XGBClassifier(
-        min_child_weight=1000, tree_method="hist", n_estimators=1, use_label_encoder=False
-    )
+    cls = xgb.XGBClassifier(min_child_weight=1000, tree_method="hist", n_estimators=1)
     cls.fit(X, y)
     assert np.all(cls.feature_importances_ == 0)
 
@@ -497,7 +492,7 @@ def dummy_objective(y_true, y_preds):
         X, y
     )
 
-    cls = xgb.XGBClassifier(use_label_encoder=False, n_estimators=1)
+    cls = xgb.XGBClassifier(n_estimators=1)
     cls.fit(X, y)
 
     is_called = [False]
@@ -923,7 +918,7 @@ def test_RFECV():
     bst = xgb.XGBClassifier(booster='gblinear', learning_rate=0.1,
                             n_estimators=10,
                             objective='binary:logistic',
-                            random_state=0, verbosity=0, use_label_encoder=False)
+                            random_state=0, verbosity=0)
     rfecv = RFECV(estimator=bst, step=1, cv=3, scoring='roc_auc')
     rfecv.fit(X, y)
 
@@ -934,7 +929,7 @@ def test_RFECV():
                             n_estimators=10,
                             objective='multi:softprob',
                             random_state=0, reg_alpha=0.001, reg_lambda=0.01,
-                            scale_pos_weight=0.5, verbosity=0, use_label_encoder=False)
+                            scale_pos_weight=0.5, verbosity=0)
     rfecv = RFECV(estimator=bst, step=1, cv=3, scoring='neg_log_loss')
     rfecv.fit(X, y)
 
@@ -943,7 +938,7 @@ def test_RFECV():
     rfecv = RFECV(estimator=reg)
     rfecv.fit(X, y)
 
-    cls = xgb.XGBClassifier(use_label_encoder=False)
+    cls = xgb.XGBClassifier()
     rfecv = RFECV(estimator=cls, step=1, cv=3,
                   scoring='neg_mean_squared_error')
     rfecv.fit(X, y)