def test_reader__split_test_and_train_data_empty(self):
"""Check split_test_and_train_data method on the empty dataset"""
reader = DSReader(dataset_empty_path)
reader.make_dictionary()
X, y = reader.vectorize()
percent = 0.7
with self.assertRaises(Exception):
X_train, y_train, X_test, y_test = reader.split_train_and_test(
X, y, percent)
def test_reader__split_test_and_train_data_zero_size(self):
"""Check split_test_and_train_data method with argument size equals to zero"""
reader = DSReader(dataset_split_path)
reader.make_dictionary()
X, y = reader.vectorize()
percent = 0
with self.assertRaises(Exception):
X_train, y_train, X_test, y_test = reader.split_train_and_test(
X, y, percent)
def test_reader__split_test_and_train_data(self):
"""Check split_test_and_train_data method"""
reader = DSReader(dataset_split_path)
reader.make_dictionary()
X, y = reader.vectorize()
percent = 0.7
X_train, y_train, X_test, y_test = reader.split_train_and_test(
X, y, percent)
self.assertEqual(X_train.shape[0], X.shape[0] * percent)
self.assertEqual(X_test.shape[0], X.shape[0] * round(1 - percent, 2))
self.assertEqual(y_train.shape[0], y.shape[0] * percent)
self.assertEqual(y_test.shape[0], y.shape[0] * round(1 - percent, 2))