Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Please add predict_proba to estimators class not in TF1.3 DNN #12119

Closed
dartdog opened this issue Aug 8, 2017 · 2 comments
Closed

Please add predict_proba to estimators class not in TF1.3 DNN #12119

dartdog opened this issue Aug 8, 2017 · 2 comments
Labels
stat:awaiting response Status - Awaiting response from author

Comments

@dartdog
Copy link

dartdog commented Aug 8, 2017

Describe the problem

Feature Request:
The TF.learn library in TF1.2 has perdict_proba for getting probability estimates for predictions it is particularly helpful for setting probability levels up or down or just measuring them.. it is a super helpful function for analysis as well as correlations can be drawn to input variables...

There are many use cases where you may want to accept a prediction with less than a 50/50 (think unbalanced classs)
and there are other cases where you want greater certainty, for some medical decisions or credit scoring..

The DNN class in TF 1.3 with the estimator does not seem to have this function. (while tf.learn dnn in TF1.2 does...)

Estimators seem to be the way of the future, but at least IMHO we need this to make that move..

@terrytangyuan
Copy link
Member

You can use tf.estimator now with e.g. predict(..., predict_keys="probabilities")

@tatatodd tatatodd added the stat:awaiting response Status - Awaiting response from author label Aug 9, 2017
@dartdog
Copy link
Author

dartdog commented Aug 9, 2017

ahh!!! Doc pointers?

@dartdog dartdog closed this as completed Aug 9, 2017
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
stat:awaiting response Status - Awaiting response from author
Projects
None yet
Development

No branches or pull requests

3 participants