Fix bug in dataset.py; fix test about split

Author: mfeurer

openml/entities/dataset.py

@@ -134,7 +134,7 @@ def get_dataset(self, target=None, include_row_id=False,
             raise ValueError("Cannot find a ndarray file for dataset %s at"
                              "location %s " % (self.name, path))
         else:
-            with open(path) as fh:
+            with open(path, "rb") as fh:
                 data, categorical, attribute_names = pickle.load(fh)
 
         to_exclude = []

openml/entities/split.py

@@ -43,10 +43,11 @@ def __eq__(self, other):
                     return False
                 else:
                     for fold in self.split[repetition]:
-                        if all(self.split[repetition][fold][0] != \
-                                other.split[repetition][fold][0]) and \
-                                all(self.split[repetition][fold][1] != \
-                                other.split[repetition][fold][1]):
+                        if np.all(self.split[repetition][fold].test != \
+                                other.split[repetition][fold].test)\
+                                and \
+                                np.all(self.split[repetition][fold].train
+                                != other.split[repetition][fold].train):
                             return False
         return True
 

tests/entities/test_split.py

@@ -43,7 +43,7 @@ def test_eq(self):
         self.assertNotEqual(split, split2)
 
         split2 = OpenMLSplit.from_arff_file(self.arff_filename)
-        split2.split[0][0] = (np.zeros((80)), np.zeros((9)))
+        split2.split[0][0] = Split(np.zeros((80)), np.zeros((9)))
         self.assertNotEqual(split, split2)
 
     def test_from_arff_file(self):