def test_inverse_raise_exception(self):
# given
mock = self.DataTransformerMock1()
p = Pipeline([mock])
# when & then
with self.assertRaises(ValueError):
p.inverse_transform(None)
def test_raises_on_bad_key(self):
# given
bad_key = 12.0
p = Pipeline([])
# when & then
with self.assertRaises(ValueError,
msg="Key must be int, str or slice"):
p[bad_key]
def test_transformers_not_modified(self):
# given
mock = self.DataTransformerMock1()
p = Pipeline([mock], copy=True)
# when
p.transform(constant_timeseries(1, 10))
# then
self.assertFalse(mock.transform_called)
def test_raises_on_non_transformers(self):
# given
input_list = list(range(10))
# when & then
with self.assertRaises(
ValueError,
msg=
"transformers should be objects deriving from BaseDataTransformer"
):
Pipeline(input_list)
def test_inverse_transform(self):
# given
data = constant_timeseries(0., 3)
transformers = [self.PlusTenTransformer(), self.TimesTwoTransformer()]
p = Pipeline(transformers)
# when
transformed = p.transform(data)
back = p.inverse_transform(transformed)
# then
self.assertEqual(data, back)
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_getitem(self):
# given
transformers = [
self.PlusTenTransformer(name="+10 transformer{}".format(i))
for i in range(0, 10)
]
p = Pipeline(transformers)
# when & then
# note : only compares string representations, since __getitem__() copies the transformers
self.assertEqual(str(p[1]._transformers), str([transformers[1]]))
self.assertEqual(str(p[4:8]._transformers), str(transformers[4:8]))
with self.assertRaises(ValueError):
p["invalid attempt"]
def test_transform_fit(self):
# given
data = constant_timeseries(value=0, length=3)
transformers = [self.DataTransformerMock1() for _ in range(10)
] + [self.DataTransformerMock2() for _ in range(10)]
p = Pipeline(transformers)
# when
_ = p.fit_transform(data)
# then
for t in transformers:
self.assertTrue(t.transform_called)
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_pipeline_n_jobs(self):
"""
Checks if the n_jobs param applied to the pipeline is changing the verbosity level in the
contained transformers.
"""
def plus_ten(x):
return x + 10
mapper = Mapper(fn=plus_ten, n_jobs=1)
mapper_inv = InvertibleMapper(fn=lambda x: x + 2,
inverse_fn=lambda x: x - 2,
n_jobs=2)
n_jobs_value = -1
pipeline = Pipeline([mapper, mapper_inv], n_jobs=n_jobs_value)
for transformer in pipeline:
self.assertEqual(transformer._n_jobs, n_jobs_value)
def test_pipeline_verbose(self):
"""
Checks if the verbose param applied to the pipeline is changing the verbosity level in the
contained transformers.
"""
def plus_ten(x):
return x + 10
mapper = Mapper(fn=plus_ten, verbose=True)
mapper_inv = InvertibleMapper(fn=lambda x: x + 2,
inverse_fn=lambda x: x - 2,
verbose=True)
verbose_value = False
pipeline = Pipeline([mapper, mapper_inv], verbose=verbose_value)
for transformer in pipeline:
self.assertEqual(transformer._verbose, verbose_value)
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)
def test_multi_ts(self):
series1 = constant_timeseries(0., 3)
series2 = constant_timeseries(1., 3)
data = [series1, series2]
mapper1 = InvertibleMapper(fn=lambda x: x + 10,
inverse_fn=lambda x: x - 10)
mapper2 = InvertibleMapper(fn=lambda x: x * 10,
inverse_fn=lambda x: x / 10)
transformers = [mapper1, mapper2]
p = Pipeline(transformers)
# when
transformed = p.transform(data)
back = p.inverse_transform(transformed)
# then
self.assertEqual(data, back)
def test_pipeline_partial_inverse(self):
series = constant_timeseries(0., 3)
def plus_ten(x):
return x + 10
mapper = Mapper(fn=plus_ten)
mapper_inv = InvertibleMapper(fn=lambda x: x + 2,
inverse_fn=lambda x: x - 2)
series_plus_ten = mapper.transform(series)
pipeline = Pipeline([mapper, mapper_inv])
transformed = pipeline.transform(series)
# should fail, since partial is False by default
with self.assertRaises(ValueError):
pipeline.inverse_transform(transformed)
back = pipeline.inverse_transform(transformed, partial=True)
# while the +/- 2 is inverted, the +10 operation is not
self.assertEqual(series_plus_ten, back)