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

Replacing model with predictions data #209

Open
caitlinrs opened this issue Jul 20, 2022 · 1 comment
Open

Replacing model with predictions data #209

caitlinrs opened this issue Jul 20, 2022 · 1 comment

Comments

@caitlinrs
Copy link

Is there a way to get the What-if Widget to work without using a model? My project does not have a model, but what I do have is prediction data. Is there a way to substitute the model with using label and prediction data instead of a model while still being able to get all the features (i.e. optimization strategy, ROC curve, PR curve, etc.)?

@jameswex
Copy link
Collaborator

There are two approaches that would work:

  1. Have the datapoints contain the prediction outputs directly with specially named fields. See the note at the bottom of https://pair-code.github.io/what-if-tool/learn/tutorials/notebooks/

  2. Write a custom prediction function that looks like the pre-calculated prediction data for each example instead of having it call the model (https://pair-code.github.io/what-if-tool/learn/tutorials/custom-prediction/)

Note that in either of these cases, some features (partial dependence plots, for example) won't work as expected as there is no way to get predictions for newly created datapoints.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants