DOC/TST: Fix Python sklearn dep

python-package/xgboost/compat.py

@@ -48,11 +48,13 @@ try:
     from sklearn.cross_validation import KFold, StratifiedKFold
     SKLEARN_INSTALLED = True
 
-    XGBKFold = KFold
-    XGBStratifiedKFold = StratifiedKFold
     XGBModelBase = BaseEstimator
     XGBRegressorBase = RegressorMixin
     XGBClassifierBase = ClassifierMixin
+
+    XGBKFold = KFold
+    XGBStratifiedKFold = StratifiedKFold
+    XGBLabelEncoder = LabelEncoder
 except ImportError:
     SKLEARN_INSTALLED = False
 
@@ -60,5 +62,7 @@ except ImportError:
     XGBModelBase = object
     XGBClassifierBase = object
     XGBRegressorBase = object
+
     XGBKFold = None
     XGBStratifiedKFold = None
+    XGBLabelEncoder = None

python-package/xgboost/sklearn.py

@@ -7,8 +7,10 @@ import numpy as np
 from .core import Booster, DMatrix, XGBoostError
 from .training import train
 
+# Do not use class names on scikit-learn directly.
+# Re-define the classes on .compat to guarantee the behavior without scikit-learn
 from .compat import (SKLEARN_INSTALLED, XGBModelBase,
-                     XGBClassifierBase, XGBRegressorBase, LabelEncoder)
+                     XGBClassifierBase, XGBRegressorBase, XGBLabelEncoder)
 
 
 def _objective_decorator(func):
@@ -398,7 +400,7 @@ class XGBClassifier(XGBModel, XGBClassifierBase):
 
         self._features_count = X.shape[1]
 
-        self._le = LabelEncoder().fit(y)
+        self._le = XGBLabelEncoder().fit(y)
         training_labels = self._le.transform(y)
 
         if sample_weight is not None:

python-package/xgboost/testing.py

@@ -0,0 +1,22 @@
+# coding: utf-8
+
+import nose
+
+from xgboost.compat import SKLEARN_INSTALLED, PANDAS_INSTALLED
+
+
+def _skip_if_no_sklearn():
+    if not SKLEARN_INSTALLED:
+        raise nose.SkipTest()
+
+
+def _skip_if_no_pandas():
+    if not PANDAS_INSTALLED:
+        raise nose.SkipTest()
+
+
+def _skip_if_no_matplotlib():
+    try:
+        import matplotlib.pyplot as plt     # noqa
+    except ImportError:
+        raise nose.SkipTest()

tests/python/test_early_stopping.py

@@ -1,15 +1,18 @@
 import xgboost as xgb
+import xgboost.testing as tm
 import numpy as np
-from sklearn.datasets import load_digits
-from sklearn.cross_validation import train_test_split
-from sklearn.metrics import mean_squared_error
 import unittest
 
 rng = np.random.RandomState(1994)
 
 
 class TestEarlyStopping(unittest.TestCase):
+
     def test_early_stopping_nonparallel(self):
+        tm._skip_if_no_sklearn()
+        from sklearn.datasets import load_digits
+        from sklearn.cross_validation import train_test_split
+
         digits = load_digits(2)
         X = digits['data']
         y = digits['target']
@@ -30,10 +33,16 @@ class TestEarlyStopping(unittest.TestCase):
         assert clf3.best_score == 1
 
     def evalerror(self, preds, dtrain):
+        tm._skip_if_no_sklearn()
+        from sklearn.metrics import mean_squared_error
+
         labels = dtrain.get_label()
         return 'rmse', mean_squared_error(labels, preds)
 
     def test_cv_early_stopping(self):
+        tm._skip_if_no_sklearn()
+        from sklearn.datasets import load_digits
+
         digits = load_digits(2)
         X = digits['data']
         y = digits['target']

tests/python/test_eval_metrics.py

@@ -1,8 +1,6 @@
 import xgboost as xgb
+import xgboost.testing as tm
 import numpy as np
-from sklearn.cross_validation import train_test_split
-from sklearn.metrics import mean_squared_error
-from sklearn.datasets import load_digits
 import unittest
 
 rng = np.random.RandomState(1337)
@@ -42,16 +40,26 @@ class TestEvalMetrics(unittest.TestCase):
         return [('error', float(sum(labels != (preds > 0.0))) / len(labels))]
 
     def evalerror_03(self, preds, dtrain):
+        tm._skip_if_no_sklearn()
+        from sklearn.metrics import mean_squared_error
+
         labels = dtrain.get_label()
         return [('rmse', mean_squared_error(labels, preds)),
                 ('error', float(sum(labels != (preds > 0.0))) / len(labels))]
 
     def evalerror_04(self, preds, dtrain):
+        tm._skip_if_no_sklearn()
+        from sklearn.metrics import mean_squared_error
+
         labels = dtrain.get_label()
         return [('error', float(sum(labels != (preds > 0.0))) / len(labels)),
                 ('rmse', mean_squared_error(labels, preds))]
 
     def test_eval_metrics(self):
+        tm._skip_if_no_sklearn()
+        from sklearn.cross_validation import train_test_split
+        from sklearn.datasets import load_digits
+
         digits = load_digits(2)
         X = digits['data']
         y = digits['target']

tests/python/test_plotting.py

@@ -1,19 +1,27 @@
 # -*- coding: utf-8 -*-
 import numpy as np
 import xgboost as xgb
+import xgboost.testing as tm
 import unittest
 
+try:
     import matplotlib
+    matplotlib.use('Agg')
     from matplotlib.axes import Axes
     from graphviz import Digraph
+except ImportError:
+    pass
+
+
+tm._skip_if_no_matplotlib()
 
-matplotlib.use('Agg')
 
 dpath = 'demo/data/'
 rng = np.random.RandomState(1994)
 
 
 class TestPlotting(unittest.TestCase):
+
     def test_plotting(self):
         bst2 = xgb.Booster(model_file='xgb.model')
 

tests/python/test_training_continuation.py

@@ -1,7 +1,6 @@
 import xgboost as xgb
+import xgboost.testing as tm
 import numpy as np
-from sklearn.metrics import mean_squared_error
-from sklearn.datasets import load_digits
 import unittest
 
 rng = np.random.RandomState(1337)
@@ -29,6 +28,10 @@ class TestTrainingContinuation(unittest.TestCase):
     }
 
     def test_training_continuation(self):
+        tm._skip_if_no_sklearn()
+        from sklearn.datasets import load_digits
+        from sklearn.metrics import mean_squared_error
+
         digits_2class = load_digits(2)
         digits_5class = load_digits(5)
 

tests/python/test_with_pandas.py

@@ -1,15 +1,26 @@
 # -*- coding: utf-8 -*-
 import numpy as np
 import xgboost as xgb
+import xgboost.testing as tm
 import unittest
+
+try:
     import pandas as pd
+except ImportError:
+    pass
+
+
+tm._skip_if_no_pandas()
+
 
 dpath = 'demo/data/'
 rng = np.random.RandomState(1994)
 
 
 class TestPandas(unittest.TestCase):
+
     def test_pandas(self):
+
         df = pd.DataFrame([[1, 2., True], [2, 3., False]], columns=['a', 'b', 'c'])
         dm = xgb.DMatrix(df, label=pd.Series([1, 2]))
         assert dm.feature_names == ['a', 'b', 'c']

tests/python/test_with_sklearn.py

@@ -1,15 +1,16 @@
+import numpy as np
 import random
 import xgboost as xgb
-import numpy as np
+import xgboost.testing as tm
-from sklearn.metrics import mean_squared_error
-from sklearn.grid_search import GridSearchCV
-from sklearn.datasets import load_iris, load_digits, load_boston
-from sklearn.cross_validation import KFold, StratifiedKFold, train_test_split
 
 rng = np.random.RandomState(1994)
 
 
 def test_binary_classification():
+    tm._skip_if_no_sklearn()
+    from sklearn.datasets import load_digits
+    from sklearn.cross_validation import KFold
+
     digits = load_digits(2)
     y = digits['target']
     X = digits['data']
@@ -24,6 +25,9 @@ def test_binary_classification():
 
 
 def test_multiclass_classification():
+    tm._skip_if_no_sklearn()
+    from sklearn.datasets import load_iris
+    from sklearn.cross_validation import KFold
 
     def check_pred(preds, labels):
         err = sum(1 for i in range(len(preds))
@@ -50,6 +54,9 @@ def test_multiclass_classification():
 
 
 def test_feature_importances():
+    tm._skip_if_no_sklearn()
+    from sklearn.datasets import load_digits
+
     digits = load_digits(2)
     y = digits['target']
     X = digits['data']
@@ -81,6 +88,11 @@ def test_feature_importances():
 
 
 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
+
     boston = load_boston()
     y = boston['target']
     X = boston['data']
@@ -102,6 +114,10 @@ def test_boston_housing_regression():
 
 
 def test_parameter_tuning():
+    tm._skip_if_no_sklearn()
+    from sklearn.grid_search import GridSearchCV
+    from sklearn.datasets import load_boston
+
     boston = load_boston()
     y = boston['target']
     X = boston['data']
@@ -114,6 +130,11 @@ def test_parameter_tuning():
 
 
 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
+
     def objective_ls(y_true, y_pred):
         grad = (y_pred - y_true)
         hess = np.ones(len(y_true))
@@ -143,6 +164,10 @@ 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
+
     def logregobj(y_true, y_pred):
         y_pred = 1.0 / (1.0 + np.exp(-y_pred))
         grad = y_pred - y_true
@@ -178,6 +203,10 @@ 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
+
     iris = load_iris()
     tr_d, te_d, tr_l, te_l = train_test_split(iris.data, iris.target, train_size=120)
 
@@ -191,6 +220,9 @@ def test_sklearn_api():
 
 
 def test_sklearn_plotting():
+    tm._skip_if_no_sklearn()
+    from sklearn.datasets import load_iris
+
     iris = load_iris()
 
     classifier = xgb.XGBClassifier()
@@ -217,6 +249,10 @@ def test_sklearn_plotting():
 
 
 def test_sklearn_nfolds_cv():
+    tm._skip_if_no_sklearn()
+    from sklearn.datasets import load_digits
+    from sklearn.cross_validation import StratifiedKFold
+
     digits = load_digits(3)
     X = digits['data']
     y = digits['target']
@@ -243,6 +279,9 @@ def test_sklearn_nfolds_cv():
 
 
 def test_split_value_histograms():
+    tm._skip_if_no_sklearn()
+    from sklearn.datasets import load_digits
+
     digits_2class = load_digits(2)
 
     X = digits_2class['data']

tests/travis/run_test.sh

@@ -45,13 +45,13 @@ if [ ${TASK} == "python_lightweight_test" ]; then
     python --version
     conda install numpy scipy nose
     python -m pip install graphviz
-    python -m nose tests/python/test_basic*.py || exit -1
+    python -m nose tests/python || exit -1
     source activate python2
     echo "-------------------------------"
     python --version
     conda install numpy scipy nose
     python -m pip install graphviz
-    python -m nose tests/python/test_basic*.py || exit -1
+    python -m nose tests/python || exit -1
     python -m pip install flake8
     flake8 --ignore E501 python-package || exit -1
     flake8 --ignore E501 tests/python || exit -1