def test_AutoWrapper_one_dimensional_transformer(column_prefix, nb_outputs,
with_get_feature_names):
np.random.seed(123)
X = np.random.randn(100, 10)
df = pd.DataFrame(X, columns=[f"NUMBER_{j}" for j in range(X.shape[1])])
df["not_a_number"] = "a"
if with_get_feature_names:
klass = DummyOnlyOneDimensionWithGetFeatureNames
else:
klass = DummyOnlyOneDimension
dummy_auto_wrapped = AutoWrapper(
klass(nb_outputs=nb_outputs),
work_on_one_column_only=True,
all_columns_at_once=False)(column_prefix=column_prefix)
Xres = dummy_auto_wrapped.fit_transform(df)
assert isinstance(Xres, pd.DataFrame)
assert (Xres.iloc[:, 0].values == df.iloc[:, 0].values).all()
assert (Xres.iloc[:,
1 + (nb_outputs - 1)].values == df.iloc[:,
1].values).all()
expected_cols = []
for col in df.columns:
add = "FEATURE" if with_get_feature_names else ""
if column_prefix is not None:
if add == "":
add = column_prefix
else:
add = column_prefix + "__" + add
if add != "":
add = "__" + add
expected_cols += [(col + add + "__%d" % d) for d in range(nb_outputs)]
assert Xres.shape == (df.shape[0], df.shape[1] * nb_outputs)
assert dummy_auto_wrapped.get_feature_names() == expected_cols
assert dummy_auto_wrapped.get_feature_names() == list(Xres.columns)
def test_AutoWrapper_fails_if_not_instance():
model = 10
with pytest.raises(TypeError):
AutoWrapper(model)
def test_AutoWrapper():
np.random.seed(123)
X = np.random.randn(100, 10)
df = pd.DataFrame(X, columns=[f"NUMBER_{j}" for j in range(X.shape[1])])
df["not_a_number"] = "a"
model = AutoWrapper(TruncatedSVD(n_components=2, random_state=123))(
columns_to_use=["NUMBER_"], regex_match=True)
Xres = model.fit_transform(df)
assert isinstance(Xres, pd.DataFrame)
assert Xres.shape[0] == df.shape[0]
model = AutoWrapper(TruncatedSVD(n_components=2, random_state=123))(
columns_to_use=["NUMBER_"], regex_match=True, column_prefix="SVD")
Xres = model.fit_transform(df)
assert isinstance(Xres, pd.DataFrame)
assert list(Xres.columns) == ["not_a_number", "SVD__0", "SVD__1"]
assert Xres.shape[0] == df.shape[0]
dummy_not_wrapped = DummyOnlyDataFrame()
with pytest.raises(TypeError):
dummy_not_wrapped.fit_transform(X)
dummy_auto_wrapped = AutoWrapper(DummyOnlyDataFrame())()
Xres = dummy_auto_wrapped.fit_transform(X)
assert isinstance(Xres, pd.DataFrame)
assert (Xres.values == X).all()
dummy_auto_wrapped = AutoWrapper(DummyOnlyDataFrame)()
Xres = dummy_auto_wrapped.fit_transform(X)
assert isinstance(Xres, pd.DataFrame)
assert (Xres.values == X).all()
dummy_not_wrapped = DummyNoDataFrame()
with pytest.raises(TypeError):
dummy_not_wrapped.fit_transform(df)
dummy_auto_wrapped = AutoWrapper(
DummyNoDataFrame,
wrapping_kwargs={"accepted_input_types": (DataTypes.NumpyArray, )})()
Xres = dummy_auto_wrapped.fit_transform(df)
assert isinstance(Xres, pd.DataFrame)