def test_score_func(self):
with pytest.raises(TypeError):
kbest = SelectKBest(score_func=10, k=10)
kbest.fit(X_class, y_reg)
with pytest.raises(TypeError):
kbest = SelectKBest(score_func="chi2", k=10)
kbest.fit(X_class, y_reg)
def test_default_score_func(self):
kbest = SelectKBest(k=10)
kbest.fit(X_reg, y_reg)
assert kbest.score_func.__name__ == f_regression.__name__
kbest = SelectKBest()
kbest.fit(X_class, y_class)
assert kbest.score_func.__name__ == f_classif.__name__
def test_default_score_func(self):
kbest = SelectKBest()
kbest.fit({"k": 10, "X": X_reg, "y": y_reg})
assert kbest.score_func.__name__ == f_regression.__name__
kbest = SelectKBest()
kbest.fit({"X": X_class, "y": y_class})
assert kbest.score_func.__name__ == f_classif.__name__
def __init__(
self,
train_df_path=None,
test_df_path=None,
steps=None,
config_file=None,
params=None,
custom_reader=None,
):
steps = [
Parser().parse_dataset,
NullValuesHandler().execute,
Encoder().encode,
HandleOutlier().handle_outliers,
Scaler().execute,
SelectKBest().fit_transform,
Split().train_test_split,
]
super().__init__(
train_df_path=train_df_path,
test_df_path=test_df_path,
steps=steps,
config_file=config_file,
params=params,
custom_reader=custom_reader,
)
def test_score_func(self):
with pytest.raises(TypeError):
kbest = SelectKBest()
kbest.fit({"score_func": 10, "k": 10, "X": X_class, "y": y_reg})
with pytest.raises(TypeError):
kbest = SelectKBest()
kbest.fit({
"score_func": "chi2",
"k": 10,
"X": X_class,
"y": y_reg
})
def test_transform_without_fit():
kbest = SelectKBest()
with pytest.raises(ValueError):
kbest.transform({"train_df": X_reg, "target_col": "target"})
def test_default_score_func(test_input, test_output):
kbest = SelectKBest()
kbest.fit(params=test_input)
assert kbest.score_func.__name__ == test_output
def test_score_func(test_input):
with pytest.raises(TypeError):
kbest = SelectKBest()
kbest.fit(params=test_input)
def test_invalid_input():
kbest = SelectKBest()
with pytest.raises(TypeError):
kbest.fit_transform({})
def test_fit_transform(params, split_size):
kbest = SelectKBest()
kbest.fit_transform(params)
assert params["train_df"].shape == (split_size, params["k"] + 1)
if "test_df" in params.keys():
assert params["test_df"].shape == (1000 - split_size, params["k"] + 1)
def test_fit_transform():
kbest = SelectKBest()
params = {"score_func": f_regression, "k": 5, "X": X_reg, "y": y_reg}
kbest.fit_transform(params)
assert params["X_best"].shape[1] == 5
def test_transform_without_fit():
kbest = SelectKBest()
with pytest.raises(ValueError):
kbest.transform(X_reg)
def test_fit_transform(self):
kbest = SelectKBest(score_func=f_regression, k=5)
X_new = kbest.fit_transform(X_reg, y_reg)
assert X_new.shape[1] == 5
