def transform(self, X):
X = X.copy()
# check that the values are non-negative for log transform
if not (X[self.variables] > 0).all().all():
vars_ = self.variables[(X[self.variables] <= 0).any()]
raise errors.InvalidModelInputError(
f"Variables contain zero or negative values, "
f"can't apply log for vars: {vars_}")
for feature in self.variables:
X[feature] = np.log(X[feature])
return X
def transform(self, X):
# encode labels
X = X.copy()
for feature in self.variables:
X[feature] = X[feature].map(self.encoder_dict_[feature])
# check if transformer introduces NaN
if X[self.variables].isnull().any().any():
null_counts = X[self.variables].isnull().any()
vars_ = {key: value for (key, value) in null_counts.items()
if value is True}
raise errors.InvalidModelInputError(
f'Categorical encoder has introduced NaN when '
f'transforming categorical variables: {vars_.keys()}')
return X