def test_invalid_window_size(self):
output_name = "moving_average"
window_size = -1
df = pd.DataFrame.from_dict({"old_name": [0, 1, 2, 3, 4, 5]})
ma_feature = MovingAverageFeature(window_size=window_size,
output_name=output_name)
with pytest.raises(ValueError):
ma_feature.fit_transform(df)
def test_random_ts_and_window_size(self, df: pd.DataFrame,
window_size: int):
output_name = "time_series"
ma_feature = MovingAverageFeature(window_size=window_size,
output_name=output_name)
df_ma = ma_feature.fit_transform(df)
expected_df_ma = self._correct_ma(df, window_size)
testing.assert_frame_equal(expected_df_ma, df_ma)
def test_correct_x_shifts(self):
horizon = 2
ts = pd.DataFrame.from_dict({"ignored": range(10)})
ts.index = np.random.random(len(ts))
features = [
MovingAverageFeature(window_size=2, output_name="mov_avg_2"),
MovingAverageFeature(window_size=5, output_name="mov_avg_5"),
ShiftFeature(shift=3, output_name="shift_3"),
]
feature_creation = FeatureCreation(horizon=horizon,
time_series_features=features)
x_shifts = feature_creation._create_x_features(ts)
expected_x_shifts = pd.DataFrame.from_dict({
"mov_avg_2": [
np.nan,
0.5,
1.5,
2.5,
3.5,
4.5,
5.5,
6.5,
7.5,
8.5,
],
"mov_avg_5": [
np.nan,
np.nan,
np.nan,
np.nan,
2,
3,
4,
5,
6,
7,
],
"shift_3": [
np.nan,
np.nan,
np.nan,
0,
1,
2,
3,
4,
5,
6,
],
})
expected_x_shifts.index = ts.index
pd.testing.assert_frame_equal(expected_x_shifts, x_shifts)
def test_positive_window_size(self):
output_name = "moving_average"
window_size = 2
df = pd.DataFrame.from_dict({"old_name": [0, 1, 2, 3, 4, 5]})
ma_feature = MovingAverageFeature(window_size=window_size,
output_name=output_name)
df_ma = ma_feature.fit_transform(df)
expected_df_ma = pd.DataFrame.from_dict(
{output_name: [np.nan, 0.5, 1.5, 2.5, 3.5, 4.5]})
testing.assert_frame_equal(expected_df_ma, df_ma)
def test_correct_fit_transform():
n_cols = 10
testing.N, testing.K = 500, n_cols
df = testing.makeTimeDataFrame(freq="MS")
output_name = "shift"
ma_feature = MovingAverageFeature(window_size=2, output_name=output_name)
fit_transform_res = ma_feature.fit_transform(df)
transform_res = ma_feature.fit(df).transform(df)
testing.assert_frame_equal(fit_transform_res, transform_res)
def test_correct_y_shifts_random_ts(self, ts, horizon):
features = [
MovingAverageFeature(window_size=2, output_name="mov_avg_2"),
MovingAverageFeature(window_size=5, output_name="mov_avg_5"),
ShiftFeature(shift=3, output_name="shift_3"),
]
feature_creation = FeatureCreation(horizon=horizon,
time_series_features=features)
x_shifts = feature_creation._create_x_features(ts)
expected_x_shifts = self._correct_x(ts, features)
pd.testing.assert_frame_equal(expected_x_shifts, x_shifts)
def test_multi_columns_window_size(self):
output_name = "moving_average"
window_size = 2
df = pd.DataFrame.from_dict({
"old_name_0": [0, 1, 2, 3, 4, 5],
"old_name_1": [7, 8, 9, 10, 11, 12]
})
ma_feature = MovingAverageFeature(window_size=window_size,
output_name=output_name)
df_ma = ma_feature.fit_transform(df)
expected_df_ma = pd.DataFrame.from_dict({
f"{output_name}_0": [np.nan, 0.5, 1.5, 2.5, 3.5, 4.5],
f"{output_name}_1": [np.nan, 7.5, 8.5, 9.5, 10.5, 11.5],
})
testing.assert_frame_equal(expected_df_ma, df_ma)
import pytest
import hypothesis.strategies as st
from hypothesis import given, settings, HealthCheck
from giottotime.feature_creation import ShiftFeature, MovingAverageFeature
from giottotime.model_selection.feature_splitters import FeatureSplitter
from giottotime.utils.hypothesis.feature_matrices import X_y_matrices
FEATURES = [
ShiftFeature(0, "0"),
ShiftFeature(1, "1"),
MovingAverageFeature(3, "3"),
]
HORIZON = 4
class TestFeatureSplitter:
def test_constructor(self):
FeatureSplitter()
@given(st.text().filter(lambda x: x != "any"))
def test_constructor_wrong_parameter(self, drop_na_mode: str):
with pytest.raises(ValueError):
FeatureSplitter(drop_na_mode)
@settings(suppress_health_check=(HealthCheck.too_slow, ))
@given(
X_y_matrices(
horizon=HORIZON,