Skip to content

Use the given estimators in get_synthetic_preds_holdout - #1067

Open
arpitjain099 wants to merge 1 commit into
uber:masterfrom
arpitjain099:fix/holdout-estimators
Open

arpitjain099 wants to merge 1 commit into
uber:masterfrom
arpitjain099:fix/holdout-estimators

Conversation

@arpitjain099

Copy link
Copy Markdown
Contributor

get_synthetic_preds_holdout accepts an estimators dict and documents it, but the body never reads it. It always builds the same eight S/T/X/R learners over LinearRegression and XGBRegressor, so a caller who passes an estimator gets those back instead, with no error. get_synthetic_preds already uses the argument, and this brings the holdout version in line with it.

The existing test passes "S Learner (LR)" and "T Learner (XGB)", which are names the hardcoded loop produces anyway, so it passes either way. The new test uses a name the defaults cannot produce, and it fails on master. Calls that pass no estimators, get_synthetic_summary_holdout included, are unaffected.

The function accepts an estimators dict and documents it, but the body never
reads it and always builds the same eight S/T/X/R learners over
LinearRegression and XGBRegressor, so a caller who passes their own estimator
gets those defaults back instead.

get_synthetic_preds already uses the argument, so mirror its if-estimators
branch here, fitting on the training split and predicting on both splits. The
existing test cannot catch this, because the two names it passes are names the
hardcoded loop produces anyway.

Co-Authored-By: Claude <noreply@anthropic.com>
Signed-off-by: Arpit Jain <arpitjain099@gmail.com>

This branch has not been deployed

No deployments
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

Successfully merging this pull request may close these issues.

1 participant