feat: probability calibration + richer classification metrics#13
Merged
Conversation
Modeling-depth batch: - CalibratorPort + SklearnCalibrator (cross-validated CalibratedClassifierCV); opt-in AutoML(calibrate=True) wraps the winning classifier after selection so served probabilities are trustworthy. Off by default (classical-first unchanged); DI-resolvable. - Evaluator now reports average_precision (PR-AUC, key on imbalanced data) and brier_score (probability quality) for binary tasks, alongside roc_auc/accuracy. Additive — no change to defaults or CV scoring. TDD on real breast_cancer data; docs note in automl.md.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Modeling-depth batch 1 (TDD, real data).
Probability calibration
CalibratorPort+SklearnCalibrator(cross-validatedCalibratedClassifierCV, isotonic).AutoML(calibrate=True)wraps the winning classifier after model selection so served probabilities are trustworthy (lending/medical risk decisions). Off by default — classical-first behaviour is unchanged; DI-resolvable viafrom_context.Richer metrics
average_precision(PR-AUC — key on imbalanced data) andbrier_score(probability quality / calibration) for binary tasks, alongsideroc_auc/accuracy. Additive: no change to defaults or CV scoring.Tests (real
breast_cancer, no fakes)calibrate=True→ winner is aCalibratedClassifierCVwith valid probabilities + low holdout Brier;calibrate=Falseunchanged.Local gates green: ruff/format/pyright(0)/12 tests + strict docs. Rebased on post-#12 main (disjoint files).