Fix to_encode option in encode_features #1123
Conversation
Codecov Report
@@ Coverage Diff @@
## main #1123 +/- ##
==========================================
- Coverage 98.37% 98.36% -0.02%
==========================================
Files 126 126
Lines 13466 13477 +11
==========================================
+ Hits 13247 13256 +9
- Misses 219 221 +2
Continue to review full report at Codecov.
|
| try: | ||
| new_X[c] = pd.to_numeric(new_X[c], errors='raise') | ||
| except (TypeError, ValueError): | ||
| pass |
There was a problem hiding this comment.
If we can't figure out why this is here/if it's still necessary, maybe we should add a warning so that we can at least tell when it does end up throwing this error instead of quietly passing
There was a problem hiding this comment.
If we do add a warning, I think it'd be good to also test the warning to increase code coverage.
There was a problem hiding this comment.
Our current tests don't cover this except block - we would need to find an example where this could occur
There was a problem hiding this comment.
In the interest of getting this fix into today's release, I'm open to doing a separate issue / PR cycle on the usefulness of this block / how we want to handle the exceptions.
Fixes #1115
encode_featureshas an optionto_encodethat allows users to specify a list of features they want encoded.encode_featureswas only partially using the info provided fromto_encode-- it was only creating encoded features from the features specified byto_encode, but it was converting all features to a numeric pandas dtype instead of just the encoded ones.