Accommodation for Imbalanced Learning Edge-Case for Transform Method

uclamii / model_tuner

A library to tune the hyperparameters of common ML models. Supports calibration and custom pipelines.

Apache License 2.0

3 stars 0 forks source link

Accommodation for Imbalanced Learning Edge-Case for Transform Method #5

Closed lshpaner closed 5 months ago

lshpaner commented 5 months ago

This pull request introduces an enhancement to the xgb_early method, specifically addressing the need to handle an edge case when using the transform method without rebalancing the evaluation set. This update ensures the method can gracefully handle scenarios where the evaluation set does not require rebalancing, thus improving the robustness and flexibility of the xgb_early implementation.

elemets commented 5 months ago

Updates on this branch:

Regression metrics added
Metrics now handled as separate optional function
XGBoost early stopping Select K Best are now compatible
XGBoost early stopping and SMOTE / ADASYN are now compatible
Imbalanced learn bug fixes
Select K Best is modularised for readability