Merge pull request #336 from amueller/remove_dataset_pickles

Remove dataset pickles

Author: mfeurer

openml/datasets/dataset.py

@@ -205,7 +205,7 @@ def get_data(self, target=None, target_dtype=int, include_row_id=False,
 
         path = self.data_pickle_file
         if not os.path.exists(path):
-            raise ValueError("Cannot find a ndarray file for dataset %s at "
+            raise ValueError("Cannot find a pickle file for dataset %s at "
                              "location %s " % (self.name, path))
         else:
             with open(path, "rb") as fh:
@@ -425,4 +425,4 @@ def _data_features_supported(self):
                 if self.features[idx].data_type not in ['numeric', 'nominal']:
                     return False
             return True
-        return True
+        return True

tests/test_datasets/test_dataset.py

@@ -1,57 +1,21 @@
-import inspect
-import unittest
-import os
-
 import numpy as np
+from scipy import sparse
 import six
 
-from openml import OpenMLDataset
+from openml.testing import TestBase
+import openml
 
 
-class OpenMLDatasetTest(unittest.TestCase):
-    # Splitting not helpful, these test's don't rely on the server and take less
-    # than 5 seconds + rebuilding the test would potentially be costly
+class OpenMLDatasetTest(TestBase):
+    _multiprocess_can_split_ = True
 
     def setUp(self):
-        # Load dataset id 1
-        __file__ = inspect.getfile(OpenMLDatasetTest)
-        self.directory = os.path.dirname(__file__)
-        self.arff_filename = os.path.join(self.directory, "..", "files",
-                                          "datasets", "2", "dataset.arff")
-        self.pickle_filename = os.path.join(self.directory, "..", "files",
-                                            "datasets", "2", "dataset.pkl")
-        self.dataset = OpenMLDataset(
-            1, "anneal", 2, "Lorem ipsum.", "arff", None, None, None,
-            "2014-04-06 23:19:24", None, "Public",
-            "http://openml.liacs.nl/files/download/2/dataset_2_anneal.ORIG.arff",
-            "class", None, None, None, None, None, None, None, None, None,
-            "939966a711925e333bf4aaadeaa71135", data_file=self.arff_filename)
-
-        self.sparse_arff_filename = os.path.join(
-            self.directory, "..", "files", "datasets", "-1", "dataset.arff")
-        self.sparse_pickle_filename = os.path.join(
-            self.directory, "..", "files", "datasets", "-1", "dataset.pkl")
-        self.sparse_dataset = OpenMLDataset(
-            -1, "dexter", -1, "Lorem ipsum.", "arff", None, None, None, None,
-            None, "Public",
-            "http://www.cs.ubc.ca/labs/beta/Projects/autoweka/datasets/dexter.zip",
-            "class", None, None, None, None, None, None, None, None, None,
-            None, data_file=self.sparse_arff_filename)
-
-    def tearDown(self):
-        for file_ in [self.pickle_filename, self.sparse_pickle_filename]:
-            os.remove(file_)
-
-    ##########################################################################
-    # Pandas
-
-    @unittest.skip("Does not work right now")
-    def test_get_arff(self):
-        rval = self.dataset.get_arff()
-        self.assertIsInstance(rval, tuple)
-        self.assertIsInstance(rval[0], np.ndarray)
-        self.assertTrue(hasattr(rval[1], '__dict__'))
-        self.assertEqual(rval[0].shape, (898, ))
+        super(OpenMLDatasetTest, self).setUp()
+        openml.config.server = self.production_server
+
+        # Load dataset id 2 - dataset 2 is interesting because it contains
+        # missing values, categorical features etc.
+        self.dataset = openml.datasets.get_dataset(2)
 
     def test_get_data(self):
         # Basic usage
@@ -69,22 +33,18 @@ def test_get_data(self):
         self.assertTrue(all([isinstance(att, six.string_types)
                              for att in attribute_names]))
 
-    def test_get_sparse_dataset(self):
-        rval = self.sparse_dataset.get_data()
-        self.assertIsInstance(rval, np.ndarray)
+    def test_get_data_with_rowid(self):
+        self.dataset.row_id_attribute = "condition"
+        rval, categorical = self.dataset.get_data(
+            include_row_id=True, return_categorical_indicator=True)
         self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual((2, 20001), rval.shape)
-        rval, categorical = self.sparse_dataset.get_data(
-            return_categorical_indicator=True)
-        self.assertIsInstance(rval, np.ndarray)
-        self.assertEqual(len(categorical), 20001)
-        self.assertTrue(all([isinstance(cat, bool) for cat in categorical]))
-        rval, attribute_names = self.sparse_dataset.get_data(
-            return_attribute_names=True)
-        self.assertIsInstance(rval, np.ndarray)
-        self.assertEqual(len(attribute_names), 20001)
-        self.assertTrue(all([isinstance(att, six.string_types)
-                             for att in attribute_names]))
+        self.assertEqual(rval.shape, (898, 39))
+        self.assertEqual(len(categorical), 39)
+        rval, categorical = self.dataset.get_data(
+            include_row_id=False, return_categorical_indicator=True)
+        self.assertEqual(rval.dtype, np.float32)
+        self.assertEqual(rval.shape, (898, 38))
+        self.assertEqual(len(categorical), 38)
 
     def test_get_data_with_target(self):
         X, y = self.dataset.get_data(target="class")
@@ -98,122 +58,127 @@ def test_get_data_with_target(self):
         self.assertNotIn("class", attribute_names)
         self.assertEqual(y.shape, (898, ))
 
+    def test_get_data_rowid_and_ignore_and_target(self):
+        self.dataset.ignore_attributes = ["condition"]
+        self.dataset.row_id_attribute = ["hardness"]
+        X, y = self.dataset.get_data(target="class", include_row_id=False,
+                                     include_ignore_attributes=False)
+        self.assertEqual(X.dtype, np.float32)
+        self.assertIn(y.dtype, [np.int32, np.int64])
+        self.assertEqual(X.shape, (898, 36))
+        X, y, categorical = self.dataset.get_data(
+            target="class", return_categorical_indicator=True)
+        self.assertEqual(len(categorical), 36)
+        self.assertListEqual(categorical, [True] * 3 + [False] + [True] * 2 + [
+            False] + [True] * 23 + [False] * 3 + [True] * 3)
+        self.assertEqual(y.shape, (898, ))
+
+    def test_get_data_with_ignore_attributes(self):
+        self.dataset.ignore_attributes = ["condition"]
+        rval = self.dataset.get_data(include_ignore_attributes=True)
+        self.assertEqual(rval.dtype, np.float32)
+        self.assertEqual(rval.shape, (898, 39))
+        rval, categorical = self.dataset.get_data(
+            include_ignore_attributes=True, return_categorical_indicator=True)
+        self.assertEqual(len(categorical), 39)
+        rval = self.dataset.get_data(include_ignore_attributes=False)
+        self.assertEqual(rval.dtype, np.float32)
+        self.assertEqual(rval.shape, (898, 38))
+        rval, categorical = self.dataset.get_data(
+            include_ignore_attributes=False, return_categorical_indicator=True)
+        self.assertEqual(len(categorical), 38)
+        # TODO test multiple ignore attributes!
+
+
+class OpenMLDatasetTestSparse(TestBase):
+    _multiprocess_can_split_ = True
+
+    def setUp(self):
+        super(OpenMLDatasetTestSparse, self).setUp()
+        openml.config.server = self.production_server
+
+        self.sparse_dataset = openml.datasets.get_dataset(4136)
+
     def test_get_sparse_dataset_with_target(self):
         X, y = self.sparse_dataset.get_data(target="class")
-        self.assertIsInstance(X, np.ndarray)
+        self.assertTrue(sparse.issparse(X))
         self.assertEqual(X.dtype, np.float32)
         self.assertIsInstance(y, np.ndarray)
         self.assertIn(y.dtype, [np.int32, np.int64])
-        self.assertEqual(X.shape, (2, 20000))
+        self.assertEqual(X.shape, (600, 20000))
         X, y, attribute_names = self.sparse_dataset.get_data(
             target="class", return_attribute_names=True)
-        self.assertIsInstance(X, np.ndarray)
+        self.assertTrue(sparse.issparse(X))
         self.assertEqual(len(attribute_names), 20000)
         self.assertNotIn("class", attribute_names)
-        self.assertEqual(y.shape, (2, ))
+        self.assertEqual(y.shape, (600, ))
 
-    def test_get_data_with_rowid(self):
-        self.dataset.row_id_attribute = "condition"
-        rval, categorical = self.dataset.get_data(
-            include_row_id=True, return_categorical_indicator=True)
-        self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual(rval.shape, (898, 39))
-        self.assertEqual(len(categorical), 39)
-        rval, categorical = self.dataset.get_data(
-            include_row_id=False, return_categorical_indicator=True)
+    def test_get_sparse_dataset(self):
+        rval = self.sparse_dataset.get_data()
+        self.assertTrue(sparse.issparse(rval))
         self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual(rval.shape, (898, 38))
-        self.assertEqual(len(categorical), 38)
-
-        # TODO this is not yet supported!
-        #rowid = ["condition", "formability"]
-        #self.dataset.row_id_attribute = rowid
-        #rval = self.dataset.get_pandas(include_row_id=False)
+        self.assertEqual((600, 20001), rval.shape)
+        rval, categorical = self.sparse_dataset.get_data(
+            return_categorical_indicator=True)
+        self.assertTrue(sparse.issparse(rval))
+        self.assertEqual(len(categorical), 20001)
+        self.assertTrue(all([isinstance(cat, bool) for cat in categorical]))
+        rval, attribute_names = self.sparse_dataset.get_data(
+            return_attribute_names=True)
+        self.assertTrue(sparse.issparse(rval))
+        self.assertEqual(len(attribute_names), 20001)
+        self.assertTrue(all([isinstance(att, six.string_types)
+                             for att in attribute_names]))
 
     def test_get_sparse_dataset_with_rowid(self):
-        self.sparse_dataset.row_id_attribute = ["a_0"]
+        self.sparse_dataset.row_id_attribute = ["V256"]
         rval, categorical = self.sparse_dataset.get_data(
             include_row_id=True, return_categorical_indicator=True)
-        self.assertIsInstance(rval, np.ndarray)
+        self.assertTrue(sparse.issparse(rval))
         self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual(rval.shape, (2, 20001))
+        self.assertEqual(rval.shape, (600, 20001))
         self.assertEqual(len(categorical), 20001)
         rval, categorical = self.sparse_dataset.get_data(
             include_row_id=False, return_categorical_indicator=True)
-        self.assertIsInstance(rval, np.ndarray)
+        self.assertTrue(sparse.issparse(rval))
         self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual(rval.shape, (2, 20000))
+        self.assertEqual(rval.shape, (600, 20000))
         self.assertEqual(len(categorical), 20000)
 
-        # TODO this is not yet supported!
-        # rowid = ["condition", "formability"]
-        #self.dataset.row_id_attribute = rowid
-        #rval = self.dataset.get_pandas(include_row_id=False)
-
-    def test_get_data_with_ignore_attributes(self):
-        self.dataset.ignore_attributes = ["condition"]
-        rval = self.dataset.get_data(include_ignore_attributes=True)
-        self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual(rval.shape, (898, 39))
-        rval, categorical = self.dataset.get_data(
-            include_ignore_attributes=True, return_categorical_indicator=True)
-        self.assertEqual(len(categorical), 39)
-        rval = self.dataset.get_data(include_ignore_attributes=False)
-        self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual(rval.shape, (898, 38))
-        rval, categorical = self.dataset.get_data(
-            include_ignore_attributes=False, return_categorical_indicator=True)
-        self.assertEqual(len(categorical), 38)
-        # TODO test multiple ignore attributes!
-
     def test_get_sparse_dataset_with_ignore_attributes(self):
-        self.sparse_dataset.ignore_attributes = ["a_0"]
+        self.sparse_dataset.ignore_attributes = ["V256"]
         rval = self.sparse_dataset.get_data(include_ignore_attributes=True)
-        self.assertIsInstance(rval, np.ndarray)
+        self.assertTrue(sparse.issparse(rval))
         self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual(rval.shape, (2, 20001))
+        self.assertEqual(rval.shape, (600, 20001))
         rval, categorical = self.sparse_dataset.get_data(
             include_ignore_attributes=True, return_categorical_indicator=True)
-        self.assertIsInstance(rval, np.ndarray)
+        self.assertTrue(sparse.issparse(rval))
         self.assertEqual(len(categorical), 20001)
         rval = self.sparse_dataset.get_data(include_ignore_attributes=False)
-        self.assertIsInstance(rval, np.ndarray)
+        self.assertTrue(sparse.issparse(rval))
         self.assertEqual(rval.dtype, np.float32)
-        self.assertEqual(rval.shape, (2, 20000))
+        self.assertEqual(rval.shape, (600, 20000))
         rval, categorical = self.sparse_dataset.get_data(
             include_ignore_attributes=False, return_categorical_indicator=True)
-        self.assertIsInstance(rval, np.ndarray)
+        self.assertTrue(sparse.issparse(rval))
         self.assertEqual(len(categorical), 20000)
         # TODO test multiple ignore attributes!
 
-    def test_get_data_rowid_and_ignore_and_target(self):
-        self.dataset.ignore_attributes = ["condition"]
-        self.dataset.row_id_attribute = ["hardness"]
-        X, y = self.dataset.get_data(target="class", include_row_id=False,
-                                     include_ignore_attributes=False)
-        self.assertEqual(X.dtype, np.float32)
-        self.assertIn(y.dtype, [np.int32, np.int64])
-        self.assertEqual(X.shape, (898, 36))
-        X, y, categorical = self.dataset.get_data(
-            target="class", return_categorical_indicator=True)
-        self.assertEqual(len(categorical), 36)
-        self.assertListEqual(categorical, [True] * 3 + [False] + [True] * 2 + [
-            False] + [True] * 23 + [False] * 3 + [True] * 3)
-        self.assertEqual(y.shape, (898, ))
-
     def test_get_sparse_dataset_rowid_and_ignore_and_target(self):
-        self.sparse_dataset.ignore_attributes = ["a_0"]
-        self.sparse_dataset.row_id_attribute = ["a_1"]
+        # TODO: re-add row_id and ignore attributes
+        self.sparse_dataset.ignore_attributes = ["V256"]
+        self.sparse_dataset.row_id_attribute = ["V512"]
         X, y = self.sparse_dataset.get_data(
             target="class", include_row_id=False,
             include_ignore_attributes=False)
-        self.assertIsInstance(X, np.ndarray)
+        self.assertTrue(sparse.issparse(X))
         self.assertEqual(X.dtype, np.float32)
         self.assertIn(y.dtype, [np.int32, np.int64])
-        self.assertEqual(X.shape, (2, 19998))
+        self.assertEqual(X.shape, (600, 19998))
         X, y, categorical = self.sparse_dataset.get_data(
             target="class", return_categorical_indicator=True)
-        self.assertIsInstance(X, np.ndarray)
+        self.assertTrue(sparse.issparse(X))
         self.assertEqual(len(categorical), 19998)
         self.assertListEqual(categorical, [False] * 19998)
-        self.assertEqual(y.shape, (2, ))
+        self.assertEqual(y.shape, (600, ))

tests/test_datasets/test_dataset_functions.py

@@ -1,13 +1,14 @@
 import unittest
 import os
-import shutil
+import os
 import sys
 
 if sys.version_info[0] >= 3:
     from unittest import mock
 else:
     import mock
 
+from oslo_concurrency import lockutils
 import scipy.sparse
 
 import openml
@@ -22,27 +23,32 @@
                                        _get_dataset_description,
                                        _get_dataset_arff,
                                        _get_dataset_features,
-                                       _get_dataset_qualities, get_dataset)
+                                       _get_dataset_qualities)
 
 
 class TestOpenMLDataset(TestBase):
     _multiprocess_can_split_ = True
 
     def setUp(self):
         super(TestOpenMLDataset, self).setUp()
-        self._remove_did1()
 
     def tearDown(self):
+        self._remove_pickle_files()
         super(TestOpenMLDataset, self).tearDown()
-        self._remove_did1()
 
-    def _remove_did1(self):
+    def _remove_pickle_files(self):
         cache_dir = self.static_cache_dir
-        did_1_dir = os.path.join(cache_dir, 'datasets', '1')
-        try:
-            shutil.rmtree(did_1_dir)
-        except:
-            pass
+        for did in ['-1', '2']:
+            with lockutils.external_lock(
+                    name='datasets.functions.get_dataset:%s' % did,
+                    lock_path=os.path.join(openml.config.get_cache_directory(), 'locks'),
+            ):
+                pickle_path = os.path.join(cache_dir, 'datasets', did,
+                                           'dataset.pkl')
+                try:
+                    os.remove(pickle_path)
+                except:
+                    pass
 
     def test__list_cached_datasets(self):
         openml.config.set_cache_directory(self.static_cache_dir)