Research Article: Continual Learning for Tabular Models in Credit Risk
Abstract
We introduce a rehearsal-efficient continual learner for GBDT-style models under covariate shift. Deployed in a sandbox credit bureau, delinquency prediction AUC improved from 0.79 to 0.83 with stable calibration.
Cite this article
Hussain, Z. & Becker, T. (2025). Research Article: Continual Learning for Tabular Models in Credit Risk. Research Explorations in Global Knowledge & Technology (REGKT), 3 (6). Retrieved from https://regkt.com/article.php?id=121&slug=continual-learning-for-tabular-models-in-credit-risk