def check_cv(pipeline,
X,
y=None,
n_folds=2,
groups=None,
split_start='before_transforms',
expected_metrics={},
**params):
cv = CV(pipeline)
if split_start == 'try_all':
len_pipeline = len(pipeline.nodes)
values_to_test = ['after_transforms', 'before_transforms']
values_to_test.extend(list(range(len_pipeline)))
values_to_test.extend(list(range(-len_pipeline, 0)))
for s in values_to_test:
graph_id = '_split_start={}'.format(str(s))
results = cv.fit(X,
y,
cv=n_folds,
groups=groups,
split_start=s,
graph_id=graph_id)
check_cv_results(cv._learner_type, results, n_folds,
expected_metrics)
else:
results = cv.fit(X,
y,
cv=n_folds,
groups=groups,
split_start=split_start,
**params)
check_cv_results(cv._learner_type, results, n_folds, expected_metrics)
return results
def test_unseen_classes(self):
# Create a dataset such that cv splits miss some of the classes
X = random_df()
y = random_series()
y[95:] = range(5)
msg = 'CV didn\'t raise Warning exception b/c of minority class issue'
with self.assertRaises(Warning, msg=msg):
cv = CV([FastLinearClassifier()])
cv.fit(X, y, cv=3)
def check_cv_with_non_defaults(self,
label_name='label',
group_id='groupid',
features='Features_1',
**params):
steps = [
ToKey(columns={
'groupid2': group_id,
'label2': label_name
}),
LightGbmRanker() << {
Role.GroupId: 'groupid2',
Role.Label: 'label2',
Role.Feature: [features]
}
]
data = self.data(label_name, group_id, features)
cv = CV(steps)
results = cv.fit(data, groups='groupid', cv=4)
check_cv_results(cv._learner_type,
results,
n_folds=4,
expected_metrics={})