Fix handling NaN values when fitting JS univariate drift
#340
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## main #340 +/- ##
==========================================
+ Coverage 82.80% 82.91% +0.11%
==========================================
Files 100 100
Lines 7554 7540 -14
Branches 1357 1351 -6
==========================================
- Hits 6255 6252 -3
+ Misses 963 960 -3
+ Partials 336 328 -8 ☔ View full report in Codecov by Sentry. |
Previously the data cleaning method operated by accepting multiple dataframes and inspecting each dataframe separetely for `NaN`'s. Depending on how the data is processed after cleaning, splitting columns into separate dataframes can be rather annoying. To avoid that this commit changes the method to accept a single dataframe and a columns argument. The columns argument specifies which column subsets should be inspected for `NaN`'s, enabling the same behaviour using a more convenient syntax.
The performance calculator for binary classification had checks in place to generate an exception if the prediction column contains nothing but `NaN`'s. This behaviour contradicts the warning functionality that is in the same functions that would should return `NaN` and issue a warning. It is also inconsistent with other calculators which do issue a warning instead of raising an error. This commit removes the errors and relies on the existing warning functionality.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Primary purpose for this PR is to fix fitting Jensen Shannon method when reference data contains
NaNvalues, described in #339.Changes
NaN's from reference data when fitting JS.InvalidArgumentExceptionaboutNaNarguments from performance calculation soNaNwill be returned instead with a warning message. This should be in line with behaviour for other calculators.Note
I've not added special handling when reference data contains only
NaNvalues. This will still result in the exception mentioned in #339 (with mention of the violating method and column name). This likely applies for most univariate drift methods, if not all of them.This seems like a different case than analysis data containing only
NaNvalues that we should still evaluate.