def _make_transformer(self) -> Transformer:
"""Build a `Transformer` to transform and inverse-transform samples in the space
Returns
-------
Transformer
Precise architecture and parameters based on :attr:`transform_` and :attr:`prior`"""
if self.transform_ == "normalize":
if self.prior == "uniform":
return Pipeline([Identity(), Normalize(self.low, self.high)])
else:
return Pipeline([
Log10(),
Normalize(np.log10(self.low), np.log10(self.high))
])
else:
if self.prior == "uniform":
return Identity()
else:
return Log10()
def _make_transformer(self) -> Transformer:
"""Build a `Transformer` to transform and inverse-transform samples in the space
Returns
-------
Transformer
`Normalize` with bounds (:attr:`low`, :attr:`high`) if :attr:`transform_` == "onehot".
Else, `Identity`"""
if self.transform_ == "normalize":
return Normalize(self.low, self.high, is_int=True)
else:
return Identity()
def _make_transformer(self) -> Transformer:
"""Build a `Transformer` to transform and inverse-transform samples in the space
Returns
-------
Transformer
`CategoricalEncoder` fit to :attr:`categories` if :attr:`transform_` == "onehot". Else,
`Identity`"""
if self.transform_ == "onehot":
t = CategoricalEncoder()
t.fit(self.categories)
return t
else:
return Identity()
def test_identity_encoder():
transformer = Identity()
X = [1, 5, 9, 9, 5, 1]
transformer.fit(X)
assert_array_equal(transformer.transform(X), X)
assert_array_equal(transformer.inverse_transform(X), X)
X = ['a', True, 1, 'a', True, 1]
transformer.fit(X)
assert_array_equal(transformer.transform(X), X)
assert_array_equal(transformer.inverse_transform(X), X)
X = ["a", "b", "c", "a", "b", "c"]
transformer.fit(X)
assert_array_equal(transformer.transform(X), X)
assert_array_equal(transformer.inverse_transform(X), X)