def test_feature_selector_category():
my_pipeline = make_pipeline(
InfraCleaner(), DomainCleaner(), AddFeatures(), FeatureSelector("category")
)
df_tr = my_pipeline.transform(df_input_start)
assert_frame_equal(df_tr, df_output_select_category)
def test_feature_selector_numeric():
my_pipeline = make_pipeline(
InfraCleaner(), DomainCleaner(), AddFeatures(), FeatureSelector(np.number)
)
df_tr = my_pipeline.transform(df_input_start)
assert_frame_equal(df_tr, df_output_select_numeric)
def test_regroup_categories_to_autre():
my_pipeline = make_pipeline(
InfraCleaner(),
DomainCleaner(),
AddFeatures(),
FeatureSelector("category"),
RegroupeCreateCategoryAutre(),
)
df_tr = my_pipeline.fit_transform(df_input_start)
assert_frame_equal(df_tr, df_output_regroup_categories)
def create_preprocessing_pipeline():
""" This methode defines the pipeline used for the technical cleaning and for constructing new features
Returns
-------
sklearn.pipeline.Pipeline
pipeline used for the preprocessing
"""
preprocessing_pipeline = make_pipeline(InfraCleaner(), DomainCleaner(),
AddFeatures())
return preprocessing_pipeline
def test_add_features():
my_pipeline = make_pipeline(InfraCleaner(), DomainCleaner(), AddFeatures())
df_tr = my_pipeline.transform(df_input_start)
assert_frame_equal(df_tr, df_output_addfeatures)
def test_pipeline():
my_pipeline = make_pipeline(InfraCleaner(), DomainCleaner())
df_tr = my_pipeline.transform(df_input_start)
assert_frame_equal(df_tr, df_output_transformer)
def test_transform():
tr = DomainCleaner()
df_tr = tr.transform(df_input)
assert_frame_equal(df_tr, df_output_transformer)
def test_correct_values():
tr = DomainCleaner()
df_tr = tr._correct_values(df_input)
assert_frame_equal(df_tr, df_output_values)
def test_correct_outliers_errors():
tr = DomainCleaner()
df_tr = tr._correct_outliers_errors(df_input)
assert_frame_equal(df_tr, df_output_outliers)