Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Bug in BaseSearchCV.inverse_transform #8344
yes. whoops. PR welcome…
On 13 Feb 2017 2:58 am, "Cédric St-Jean" ***@***.***> wrote: The delegating code <https://github.com/scikit-learn/scikit-learn/blob/e5ceda88f2a24b3dd4f9a94404828f982cdf52ad/sklearn/utils/validation.py#L650> for inverse_transform is def inverse_transform(self, Xt): self._check_is_fitted('inverse_transform') return self.best_estimator_.transform(Xt) Unless I'm mistaken, this should be .inverse_transform(Xt) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#8344>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz69wCUxqHeSL4gqzwXwKAv5wLfL05ks5rbyw_gaJpZM4L-hsP> .
referenced this issue
Feb 13, 2017
no, inverse_transform can only occasionally be perfect. consider the case of feature selection. Besides, grid search is untested by common tests atm…
On 14 Feb 2017 9:02 pm, "Loïc Estève" ***@***.***> wrote: Just curious, don't we have estimator checks that make sure that estimator.inverse_transform(transform(X)) == X ? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#8344 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz69svkP01odPTDp2NKG_OSstvI4J3ks5rcXu-gaJpZM4L-hsP> .