Fix #3730: scikit-learn 0.20 compatibility fix (#3731)

* Fix #3730: scikit-learn 0.20 compatibility fix

sklearn.cross_validation has been removed from scikit-learn 0.20,
so replace it with sklearn.model_selection

* Display test names for Python tests for clarity

tests/python/test_with_sklearn.py

@@ -22,21 +22,13 @@ class TemporaryDirectory(object):
 def test_binary_classification():
     tm._skip_if_no_sklearn()
     from sklearn.datasets import load_digits
-    try:
-        from sklearn.model_selection import KFold
-    except:
-        from sklearn.cross_validation import KFold
+    from sklearn.model_selection import KFold
 
     digits = load_digits(2)
     y = digits['target']
     X = digits['data']
-    try:
-        kf = KFold(y.shape[0], n_folds=2, shuffle=True, random_state=rng)
-    except TypeError:  # sklearn.model_selection.KFold uses n_split
-        kf = KFold(
-            n_splits=2, shuffle=True, random_state=rng
-        ).split(np.arange(y.shape[0]))
-    for train_index, test_index in kf:
+    kf = KFold(n_splits=2, shuffle=True, random_state=rng)
+    for train_index, test_index in kf.split(X, y):
         xgb_model = xgb.XGBClassifier().fit(X[train_index], y[train_index])
         preds = xgb_model.predict(X[test_index])
         labels = y[test_index]
@@ -48,10 +40,7 @@ def test_binary_classification():
 def test_multiclass_classification():
     tm._skip_if_no_sklearn()
     from sklearn.datasets import load_iris
-    try:
-        from sklearn.cross_validation import KFold
-    except:
-        from sklearn.model_selection import KFold
+    from sklearn.model_selection import KFold
 
     def check_pred(preds, labels, output_margin):
         if output_margin:
@@ -65,8 +54,8 @@ def test_multiclass_classification():
     iris = load_iris()
     y = iris['target']
     X = iris['data']
-    kf = KFold(y.shape[0], n_folds=2, shuffle=True, random_state=rng)
-    for train_index, test_index in kf:
+    kf = KFold(n_splits=2, shuffle=True, random_state=rng)
+    for train_index, test_index in kf.split(X, y):
         xgb_model = xgb.XGBClassifier().fit(X[train_index], y[train_index])
         preds = xgb_model.predict(X[test_index])
         # test other params in XGBClassifier().fit
@@ -149,13 +138,13 @@ def test_boston_housing_regression():
     tm._skip_if_no_sklearn()
     from sklearn.metrics import mean_squared_error
     from sklearn.datasets import load_boston
-    from sklearn.cross_validation import KFold
+    from sklearn.model_selection import KFold
 
     boston = load_boston()
     y = boston['target']
     X = boston['data']
-    kf = KFold(y.shape[0], n_folds=2, shuffle=True, random_state=rng)
-    for train_index, test_index in kf:
+    kf = KFold(n_splits=2, shuffle=True, random_state=rng)
+    for train_index, test_index in kf.split(X, y):
         xgb_model = xgb.XGBRegressor().fit(X[train_index], y[train_index])
 
         preds = xgb_model.predict(X[test_index])
@@ -173,7 +162,7 @@ def test_boston_housing_regression():
 
 def test_parameter_tuning():
     tm._skip_if_no_sklearn()
-    from sklearn.grid_search import GridSearchCV
+    from sklearn.model_selection import GridSearchCV
     from sklearn.datasets import load_boston
 
     boston = load_boston()
@@ -181,7 +170,8 @@ def test_parameter_tuning():
     X = boston['data']
     xgb_model = xgb.XGBRegressor()
     clf = GridSearchCV(xgb_model, {'max_depth': [2, 4, 6],
-                                   'n_estimators': [50, 100, 200]}, verbose=1)
+                                   'n_estimators': [50, 100, 200]},
+                       cv=3, verbose=1, iid=True)
     clf.fit(X, y)
     assert clf.best_score_ < 0.7
     assert clf.best_params_ == {'n_estimators': 100, 'max_depth': 4}
@@ -191,7 +181,7 @@ def test_regression_with_custom_objective():
     tm._skip_if_no_sklearn()
     from sklearn.metrics import mean_squared_error
     from sklearn.datasets import load_boston
-    from sklearn.cross_validation import KFold
+    from sklearn.model_selection import KFold
 
     def objective_ls(y_true, y_pred):
         grad = (y_pred - y_true)
@@ -201,8 +191,8 @@ def test_regression_with_custom_objective():
     boston = load_boston()
     y = boston['target']
     X = boston['data']
-    kf = KFold(y.shape[0], n_folds=2, shuffle=True, random_state=rng)
-    for train_index, test_index in kf:
+    kf = KFold(n_splits=2, shuffle=True, random_state=rng)
+    for train_index, test_index in kf.split(X, y):
         xgb_model = xgb.XGBRegressor(objective=objective_ls).fit(
             X[train_index], y[train_index]
         )
@@ -224,7 +214,7 @@ def test_regression_with_custom_objective():
 def test_classification_with_custom_objective():
     tm._skip_if_no_sklearn()
     from sklearn.datasets import load_digits
-    from sklearn.cross_validation import KFold
+    from sklearn.model_selection import KFold
 
     def logregobj(y_true, y_pred):
         y_pred = 1.0 / (1.0 + np.exp(-y_pred))
@@ -235,8 +225,8 @@ def test_classification_with_custom_objective():
     digits = load_digits(2)
     y = digits['target']
     X = digits['data']
-    kf = KFold(y.shape[0], n_folds=2, shuffle=True, random_state=rng)
-    for train_index, test_index in kf:
+    kf = KFold(n_splits=2, shuffle=True, random_state=rng)
+    for train_index, test_index in kf.split(X, y):
         xgb_model = xgb.XGBClassifier(objective=logregobj)
         xgb_model.fit(X[train_index], y[train_index])
         preds = xgb_model.predict(X[test_index])
@@ -263,10 +253,11 @@ def test_classification_with_custom_objective():
 def test_sklearn_api():
     tm._skip_if_no_sklearn()
     from sklearn.datasets import load_iris
-    from sklearn.cross_validation import train_test_split
+    from sklearn.model_selection import train_test_split
 
     iris = load_iris()
-    tr_d, te_d, tr_l, te_l = train_test_split(iris.data, iris.target, train_size=120)
+    tr_d, te_d, tr_l, te_l = train_test_split(iris.data, iris.target,
+                                              train_size=120, test_size=0.2)
 
     classifier = xgb.XGBClassifier(booster='gbtree', n_estimators=10)
     classifier.fit(tr_d, tr_l)
@@ -280,7 +271,7 @@ def test_sklearn_api():
 def test_sklearn_api_gblinear():
     tm._skip_if_no_sklearn()
     from sklearn.datasets import load_iris
-    from sklearn.cross_validation import train_test_split
+    from sklearn.model_selection import train_test_split
 
     iris = load_iris()
     tr_d, te_d, tr_l, te_l = train_test_split(iris.data, iris.target, train_size=120)
@@ -514,23 +505,15 @@ def test_validation_weights_xgbclassifier():
 def test_save_load_model():
     tm._skip_if_no_sklearn()
     from sklearn.datasets import load_digits
-    try:
-        from sklearn.model_selection import KFold
-    except:
-        from sklearn.cross_validation import KFold
+    from sklearn.model_selection import KFold
 
     digits = load_digits(2)
     y = digits['target']
     X = digits['data']
-    try:
-        kf = KFold(y.shape[0], n_folds=2, shuffle=True, random_state=rng)
-    except TypeError:  # sklearn.model_selection.KFold uses n_split
-        kf = KFold(
-            n_splits=2, shuffle=True, random_state=rng
-        ).split(np.arange(y.shape[0]))
+    kf = KFold(n_splits=2, shuffle=True, random_state=rng)
     with TemporaryDirectory() as tempdir:
         model_path = os.path.join(tempdir, 'digits.model')
-        for train_index, test_index in kf:
+        for train_index, test_index in kf.split(X, y):
             xgb_model = xgb.XGBClassifier().fit(X[train_index], y[train_index])
             xgb_model.save_model(model_path)
             xgb_model = xgb.XGBModel()