Fix mean Tweedie deviance for power outside 0, 1 and 2 - #340
Merged
Conversation
The deviance formula was split across lines with the minus and plus operators at the beginning of each line. Inside parentheses Elixir parses those as separate expressions in a block, so only the last term was returned. The deviance was wrong for any power other than 0, 1 and 2, and could even be negative. This also affected d2_tweedie_score. Joins the expression into a single one and adds tests validated against scikit-learn for power 1.5, 3 and -1.
josevalim
approved these changes
Jul 15, 2026
Contributor
|
💚 💙 💜 💛 ❤️ |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
I was evaluating a regression model using
mean_tweedie_deviancewith power 1.5 (compound Poisson-gamma) and the deviance looked way off compared to what sklearn gave for the same predictions. Traced it down to the formula for power outside 0, 1 and 2: it was split across multiple lines with the minus and plus operators at the start of each line, which Elixir parses as separate expressions inside parentheses rather than a continued one, so only the last term was actually being returned. Same issue affected d2_tweedie_score.Joined it into a single expression and added tests comparing against sklearn for power 1.5, 3 and -1, plus a check that deviance is 0 for perfect predictions.