class TestDora(unittest.TestCase):
def setUp(self):
self.dora = Dora()
self.dora.configure(output = 'A', data = './spec_data.csv')
def test_configure(self):
data = pd.read_csv('./spec_data.csv')
self.assertEqual(self.dora.output, 'A')
self.assertTrue(self.dora.data.equals(data))
def test_remove_feature(self):
self.dora.remove_feature('useless_feature')
self.assertFalse('useless_feature' in self.dora.data.columns)
def test_extract_feature(self):
self.dora.extract_feature(
'useless_feature',
'another_useless_feature',
lambda x: x * 2
)
actual_column = list(self.dora.data['another_useless_feature'])
expected_column = [2, 2, 2]
self.assertEqual(actual_column, expected_column)
def test_impute_missing_values(self):
del self.dora.data['D']
self.dora.impute_missing_values()
actual_column = list(self.dora.data['B'])
expected_column = [2.0, 5.0, 8.0]
self.assertEqual(actual_column, expected_column)
def test_scale_input_values(self):
del self.dora.data['D'], self.dora.data['B']
self.dora.scale_input_values()
actual_column = list(self.dora.data['C'])
expected_column = [-1.224745, 0.0, 1.224745]
pairwise_diffs = map(
lambda actual, expected: abs(actual - expected),
actual_column,
expected_column
)
total_diff = sum(pairwise_diffs)
self.assertAlmostEqual(total_diff, 0, places = 6)
def test_extract_ordinal_feature(self):
self.dora.extract_ordinal_feature('D')
features = self.dora.data.columns
self.assertTrue('D=left' in features and 'D=right' in features)
def test_input_columns(self):
actual_input_columns = list(self.dora.input_columns())
expected_input_columns = list(self.dora.data.columns)
expected_input_columns.remove(self.dora.output)
self.assertEqual(actual_input_columns, expected_input_columns)
def test_logs(self):
self.dora.extract_ordinal_feature('D')
self.dora.impute_missing_values()
self.dora.scale_input_values()
actual_logs = self.dora.logs
expected_logs = [
"self.extract_ordinal_feature('D')",
'self.impute_missing_values()',
'self.scale_input_values()'
]
self.assertEqual(actual_logs, expected_logs)
def test_snapshots(self):
self.dora.snapshot('start')
self.dora.extract_ordinal_feature('D')
self.dora.use_snapshot('start')
self.assertEqual(self.dora.logs, [])
self.assertTrue(self.dora.data.equals(self.dora.initial_data))
def setUp(self): self.dora = Dora() self.dora.configure(output = 'A', data = './spec_data.csv')