def test_classifier_gridsearch(cls):
pipe = DebugPipeline([("ovrc", cls(LinearSVC(random_state=0, tol=0.1)))])
Cs = [0.1, 0.5, 0.8]
cv = GridSearchCV(pipe, {"ovrc__estimator__C": Cs})
cv.fit(IRIS.data, IRIS.target)
best_C = cv.best_estimator_.get_params()["ovrc__estimator__C"]
assert best_C in Cs
def test_time_in_logs_when_log_callback_is_default(caplog, steps):
pipe = DebugPipeline(steps, log_callback='default')
caplog.clear()
with caplog.at_level(logging.INFO):
pipe.fit(IRIS.data, IRIS.target)
assert caplog.text, f'Log should be none empty: {caplog.text}'
assert f'time=' in caplog.text, f'"time=" should be in: {caplog.text}'
assert caplog.text.count('time') == (len(pipe.steps) - 1), \
f'"time" should be {len(pipe.steps) - 1} times in {caplog.text}'
def test_nbytes_in_logs_when_log_callback_is_custom(caplog, steps):
pipe = DebugPipeline(steps, log_callback=custom_log_callback)
caplog.clear()
with caplog.at_level(logging.INFO):
pipe.fit(IRIS.data, IRIS.target)
assert caplog.text, f'Log should be none empty: {caplog.text}'
assert 'nbytes=' in caplog.text, f'"nbytes=" should be in: {caplog.text}'
assert caplog.text.count('nbytes=') == (len(pipe.steps) - 1), \
f'"nbytes=" should be {len(pipe.steps) - 1} times in {caplog.text}'
def test_output_shape_in_logs_when_log_callback_is_default(caplog, steps):
pipe = DebugPipeline(steps, log_callback='default')
caplog.clear()
with caplog.at_level(logging.INFO):
pipe.fit(IRIS.data, IRIS.target)
assert caplog.text, f'Log should be none empty: {caplog.text}'
shape_str = f'shape={IRIS.data.shape}'
assert shape_str in caplog.text, f'"{shape_str}" should be in {caplog.text}'
assert caplog.text.count(shape_str) == (len(pipe.steps) - 1), \
f'"{shape_str}" should be {len(pipe.steps) - 1} times in {caplog.text}'
def test_step_name_in_logs_when_log_callback_is_default(caplog, steps):
pipe = DebugPipeline(steps, log_callback='default')
caplog.clear()
with caplog.at_level(logging.INFO):
pipe.fit(IRIS.data, IRIS.target)
assert caplog.text, f'Log should be none empty: {caplog.text}'
for _, step in pipe.steps[:-1]:
assert str(step) in caplog.text, f'{step} should be in: {caplog.text}'
assert caplog.text.count(str(step)) == 1, \
f'{step} should be once in {caplog.text}'
def test_feature_union(caplog, steps):
pipe_w_default_log_callback = DebugPipeline(steps, log_callback='default')
pipe_w_custom_log_callback = DebugPipeline(
steps, log_callback=custom_log_callback)
pipe_union = FeatureUnion([
('pipe_w_default_log_callback', pipe_w_default_log_callback),
('pipe_w_custom_log_callback', pipe_w_custom_log_callback),
])
caplog.clear()
with caplog.at_level(logging.INFO):
pipe_union.fit(IRIS.data, IRIS.target)
assert caplog.text, f'Log should be none empty: {caplog.text}'
for pipe in [pipe_w_default_log_callback, pipe_w_custom_log_callback]:
for _, step in pipe.steps[:-1]:
assert str(step) in caplog.text, \
f'{step} should be in: {caplog.text}'
assert caplog.text.count(str(step)) == 2, \
f'{step} should be once in {caplog.text}'
def food_recognition_model(**params):
pca = PCA(512, whiten=True, random_state=0)
return DebugPipeline(steps=[("scale", StandardScaler()), ("pca", pca),
("print_preprocess_stats",
PrintPreprocessStats(pca)),
("model",
GaussianMixture(n_components=4,
covariance_type='full',
random_state=0,
max_iter=int(1e7)))],
log_callback='default').set_params(**params)
def test_no_logs_when_log_callback_is_None(caplog, steps):
pipe = DebugPipeline(steps, log_callback=None)
caplog.clear()
with caplog.at_level(logging.INFO):
pipe.fit(IRIS.data, IRIS.target)
assert not caplog.text, f'Log should be empty: {caplog.text}'