def test_two_numeric_one_feature_passed(self):
"""
transformation should be performed on y
"""
df = pd.DataFrame({'x': [1, 2, 3], 'y': [1, 2, 3]})
df_transformed_correct = pd.DataFrame({'x': [0, 0, 0], 'y': [1, 2, 3]})
bucket = Bucketizer(bins=1, features=['x'])
df_transformed = bucket.fit_transform(df)
self.compare_DataFrame(df_transformed, df_transformed_correct)
def test_two_numeric_all_numeric(self):
"""
transformation should be performed on y
"""
df = pd.DataFrame({'x': [1, 2, 3], 'y': [1, 2, 3]})
df_transformed_correct = pd.DataFrame({'x': [0, 0, 0], 'y': [0, 0, 0]})
bucket = Bucketizer(bins=1, bin_numeric=True)
df_transformed = bucket.fit_transform(df)
self.compare_DataFrame(df_transformed, df_transformed_correct)
def test_one_numeric_feature_prefix(self):
"""
transformation should be performed on y
"""
df = pd.DataFrame({'x': [1, 2, 3], 'y': [1, 2, 3]})
df_transformed_correct = pd.DataFrame({'x': [0, 0, 0], 'y': [1, 2, 3]})
bucket = Bucketizer(bins=1, bin_numeric=False, prefix='x')
df_transformed = bucket.fit_transform(df)
self.compare_DataFrame(df_transformed, df_transformed_correct)
def test_one_numeric_feature_no_transformation(self):
"""
no transformation should be performed
"""
df = pd.DataFrame({'x': [1, 2, 3]})
df_transformed_correct = pd.DataFrame({'x': [1, 2, 3]})
bucket = Bucketizer()
df_transformed = bucket.fit_transform(df)
self.compare_DataFrame(df_transformed, df_transformed_correct)
def test_no_numeric_feature(self):
"""
no transformation should be performed
"""
df = pd.DataFrame({'x': [np.NaN, 'b', 'c']})
df_transformed_correct = pd.DataFrame({'x': [np.NaN, 'b', 'c']})
bucket = Bucketizer(features=['x'])
df_transformed = bucket.fit_transform(df)
self.compare_DataFrame(df_transformed, df_transformed_correct)
def test_no_replace(self):
"""
transformation should be performed on y
"""
df = pd.DataFrame({'x': [1, 2, 3]})
df_transformed_correct = pd.DataFrame({
'x': [1, 2, 3],
'x_binned': [0, 0, 0]
})
bucket = Bucketizer(bins=1, replace=False, bin_numeric=True)
df_transformed = bucket.fit_transform(df)
self.compare_DataFrame(df_transformed, df_transformed_correct)
def test_empty_init(self):
"""
init object with default parameters
"""
bucket = Bucketizer()
assert bucket.features == []
assert bucket.bins == 2
assert bucket.replace
assert not bucket.bin_numeric
assert not bucket.prefix