def test_fit(self):
# given
data = constant_timeseries(0, 3)
transformers = [self.DataTransformerMock1() for _ in range(10)
] + [self.DataTransformerMock2() for _ in range(10)]
p = Pipeline(transformers)
# when
p.fit(data)
# then
for i in range(10):
self.assertFalse(transformers[i].fit_called)
for i in range(10, 20):
self.assertTrue(transformers[i].fit_called)
def test_transform(self):
# given
mock1 = self.DataTransformerMock1()
mock2 = self.DataTransformerMock2()
data = constant_timeseries(0, 3)
transformers = [mock1] * 10 + [mock2] * 10
p = Pipeline(transformers)
# when
p.fit(data)
transformed = p.transform(data)
# then
self.assertEqual(63, len(transformed))
self.assertEqual([0] * 3 + [1] * 30 + [2] * 30,
list(transformed.values()))
for t in transformers:
self.assertTrue(t.transform_called)
self.assertFalse(t.inverse_transform_called)
def test_fit_skips_superfluous_transforms(self):
# given
data = constant_timeseries(0, 100)
transformers = [self.DataTransformerMock1() for _ in range(10)]\
+ [self.DataTransformerMock2()]\
+ [self. DataTransformerMock1() for _ in range(10)]
p = Pipeline(transformers)
# when
p.fit(data)
# then
for i in range(10):
self.assertTrue(transformers[i].transform_called)
self.assertTrue(transformers[10].fit_called)
self.assertFalse(transformers[10].transform_called)
for i in range(11, 21):
self.assertFalse(transformers[i].transform_called)