def test_naming(self, time_series: pd.DataFrame):
exogenous = Exogenous()
transformed_time_series = exogenous.fit_transform(time_series)
expected_columns = [
f"{column_name}__Exogenous" for column_name in time_series.columns
]
assert expected_columns == list(transformed_time_series.columns)
def test_correct_nearest_pad_method(self):
method = "nearest"
exog = pd.DataFrame.from_dict({"x_1": [0, 1, 2, 3]})
exog.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 2, 1),
pd.Timestamp(2000, 3, 1),
pd.Timestamp(2000, 4, 1),
]
df = pd.DataFrame.from_dict({"x_2": [10, 11, 12, 13]})
df.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 1, 2),
pd.Timestamp(2000, 1, 3),
pd.Timestamp(2000, 1, 29),
]
exog_feature = Exogenous(exogenous_time_series=exog, method=method)
new_exog_feature = exog_feature.fit_transform(df)
expected_exog = pd.DataFrame.from_dict({"output_name": [0, 0, 0, 1]})
expected_exog.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 1, 2),
pd.Timestamp(2000, 1, 3),
pd.Timestamp(2000, 1, 29),
]
testing.assert_frame_equal(expected_exog, new_exog_feature)
def test_correct_exog_none_method(self):
method = None
exog = pd.DataFrame.from_dict({"x0": [0, 1, 2, 3]})
exog.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 2, 1),
pd.Timestamp(2000, 3, 1),
pd.Timestamp(2000, 4, 1),
]
df = pd.DataFrame.from_dict({"x1": [10, 11, 12, 13]})
df.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 1, 2),
pd.Timestamp(2000, 1, 3),
pd.Timestamp(2000, 1, 4),
]
exog_feature = Exogenous(exogenous_time_series=exog, method=method)
feature_name = exog_feature.__class__.__name__
new_exog_feature = exog_feature.fit_transform(df)
expected_exog = pd.DataFrame.from_dict(
{f"x0__{feature_name}": [0, np.nan, np.nan, np.nan]})
expected_exog.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 1, 2),
pd.Timestamp(2000, 1, 3),
pd.Timestamp(2000, 1, 4),
]
testing.assert_frame_equal(expected_exog, new_exog_feature)
def test_correct_multi_columns_exog(self):
output_name = "exog"
exog = pd.DataFrame.from_dict({
"x_0": [0, 1, 2, 3],
"x_1": [5, 6, 7, 8]
})
exog.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 2, 1),
pd.Timestamp(2000, 3, 1),
pd.Timestamp(2000, 4, 1),
]
df = pd.DataFrame.from_dict({"x_2": [10, 11, 12, 13]})
df.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 1, 2),
pd.Timestamp(2000, 1, 3),
pd.Timestamp(2000, 1, 29),
]
exog_feature = Exogenous(exogenous_time_series=exog)
new_exog_feature = exog_feature.fit_transform(df)
expected_exog = pd.DataFrame.from_dict({
f"{output_name}_0": [0, np.nan, np.nan, np.nan],
f"{output_name}_1": [5, np.nan, np.nan, np.nan],
})
expected_exog.index = [
pd.Timestamp(2000, 1, 1),
pd.Timestamp(2000, 1, 2),
pd.Timestamp(2000, 1, 3),
pd.Timestamp(2000, 1, 29),
]
testing.assert_frame_equal(expected_exog, new_exog_feature)
def test_multiple_columns(self, time_series: pd.DataFrame):
exogenous = Exogenous()
transformed_time_series = exogenous.fit_transform(time_series)
transformed_time_series.columns = ["A", "B"]
assert_frame_equal(transformed_time_series, time_series, check_names=False)