pytest tests/python fails if no pandas installed (#4620)

* _maybe_pandas_xxx should return their arguments unchanged if no pandas installed

* Tests should not assume pandas is installed

* Mark tests which require pandas as such

python-package/xgboost/core.py

@@ -226,7 +226,7 @@ PANDAS_DTYPE_MAPPER = {'int8': 'int', 'int16': 'int', 'int32': 'int', 'int64': '
 def _maybe_pandas_data(data, feature_names, feature_types):
     """ Extract internal data from pd.DataFrame for DMatrix data """
 
-    if not isinstance(data, DataFrame):
+    if not (PANDAS_INSTALLED and isinstance(data, DataFrame)):
         return data, feature_names, feature_types
 
     data_dtypes = data.dtypes
@@ -258,7 +258,7 @@ def _maybe_pandas_data(data, feature_names, feature_types):
 def _maybe_pandas_label(label):
     """ Extract internal data from pd.DataFrame for DMatrix label """
 
-    if isinstance(label, DataFrame):
+    if PANDAS_INSTALLED and isinstance(label, DataFrame):
         if len(label.columns) > 1:
             raise ValueError('DataFrame for label cannot have multiple columns')
 

tests/python/test_early_stopping.py

@@ -44,6 +44,11 @@ class TestEarlyStopping(unittest.TestCase):
         labels = dtrain.get_label()
         return 'rmse', mean_squared_error(labels, preds)
 
+    @staticmethod
+    def assert_metrics_length(cv, expected_length):
+        for key, value in cv.items():
+          assert len(value) == expected_length
+
     @pytest.mark.skipif(**tm.no_sklearn())
     def test_cv_early_stopping(self):
         from sklearn.datasets import load_digits
@@ -57,21 +62,21 @@ class TestEarlyStopping(unittest.TestCase):
 
         cv = xgb.cv(params, dm, num_boost_round=10, nfold=10,
                     early_stopping_rounds=10)
-        assert cv.shape[0] == 10
+        self.assert_metrics_length(cv, 10)
         cv = xgb.cv(params, dm, num_boost_round=10, nfold=10,
                     early_stopping_rounds=5)
-        assert cv.shape[0] == 3
+        self.assert_metrics_length(cv, 3)
         cv = xgb.cv(params, dm, num_boost_round=10, nfold=10,
                     early_stopping_rounds=1)
-        assert cv.shape[0] == 1
+        self.assert_metrics_length(cv, 1)
 
         cv = xgb.cv(params, dm, num_boost_round=10, nfold=10,
                     feval=self.evalerror, early_stopping_rounds=10)
-        assert cv.shape[0] == 10
+        self.assert_metrics_length(cv, 10)
         cv = xgb.cv(params, dm, num_boost_round=10, nfold=10,
                     feval=self.evalerror, early_stopping_rounds=1)
-        assert cv.shape[0] == 5
+        self.assert_metrics_length(cv, 5)
         cv = xgb.cv(params, dm, num_boost_round=10, nfold=10,
                     feval=self.evalerror, maximize=True,
                     early_stopping_rounds=1)
-        assert cv.shape[0] == 1
+        self.assert_metrics_length(cv, 1)

tests/python/test_with_sklearn.py

@@ -108,6 +108,7 @@ def test_ranking():
     np.testing.assert_almost_equal(pred, pred_orig)
 
 
+@pytest.mark.skipif(**tm.no_pandas())
 def test_feature_importances_weight():
     from sklearn.datasets import load_digits
 
@@ -140,6 +141,7 @@ def test_feature_importances_weight():
     np.testing.assert_almost_equal(xgb_model.feature_importances_, exp)
 
 
+@pytest.mark.skipif(**tm.no_pandas())
 def test_feature_importances_gain():
     from sklearn.datasets import load_digits
 
@@ -368,6 +370,7 @@ def test_sklearn_plotting():
     assert isinstance(ax, Axes)
 
 
+@pytest.mark.skipif(**tm.no_pandas())
 def test_sklearn_nfolds_cv():
     from sklearn.datasets import load_digits
     from sklearn.model_selection import StratifiedKFold
@@ -390,15 +393,16 @@ def test_sklearn_nfolds_cv():
     nfolds = 5
     skf = StratifiedKFold(n_splits=nfolds, shuffle=True, random_state=seed)
 
-    cv1 = xgb.cv(params, dm, num_boost_round=10, nfold=nfolds, seed=seed)
+    cv1 = xgb.cv(params, dm, num_boost_round=10, nfold=nfolds, seed=seed, as_pandas=True)
     cv2 = xgb.cv(params, dm, num_boost_round=10, nfold=nfolds,
-                 folds=skf, seed=seed)
+                 folds=skf, seed=seed, as_pandas=True)
     cv3 = xgb.cv(params, dm, num_boost_round=10, nfold=nfolds,
-                 stratified=True, seed=seed)
+                 stratified=True, seed=seed, as_pandas=True)
     assert cv1.shape[0] == cv2.shape[0] and cv2.shape[0] == cv3.shape[0]
     assert cv2.iloc[-1, 0] == cv3.iloc[-1, 0]
 
 
+@pytest.mark.skipif(**tm.no_pandas())
 def test_split_value_histograms():
     from sklearn.datasets import load_digits