Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.19663 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908848593305600 |
|---|---|
| author | Schutte, Willem D. Pretorius, Charl Smit, Neill van der Merwe, Leandra Maxwell, Robert |
| author_facet | Schutte, Willem D. Pretorius, Charl Smit, Neill van der Merwe, Leandra Maxwell, Robert |
| contents | In this paper, we study how class imbalance, typical of low-default credit portfolios, affects the performance of logistic regression models. Using a simulation study with controlled data-generating mechanisms, we vary (i) the level of class imbalance and (ii) the strength of association between the predictors and the response. The results show that, for a given strength of association, achievable classification accuracy deteriorates markedly as the event rate decreases, and the optimal classification cut-off shifts with the level of imbalance. In contrast, the Gini coefficient is comparatively stable with respect to class imbalance once sample sizes are sufficiently large, even when classification accuracy is strongly affected. As a practical guideline, we summarise attainable classification performance as a function of the event rate and strength of association between the predictors and the response. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_19663 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | The impact of class imbalance in logistic regression models for low-default portfolios in credit risk Schutte, Willem D. Pretorius, Charl Smit, Neill van der Merwe, Leandra Maxwell, Robert Risk Management Computation In this paper, we study how class imbalance, typical of low-default credit portfolios, affects the performance of logistic regression models. Using a simulation study with controlled data-generating mechanisms, we vary (i) the level of class imbalance and (ii) the strength of association between the predictors and the response. The results show that, for a given strength of association, achievable classification accuracy deteriorates markedly as the event rate decreases, and the optimal classification cut-off shifts with the level of imbalance. In contrast, the Gini coefficient is comparatively stable with respect to class imbalance once sample sizes are sufficiently large, even when classification accuracy is strongly affected. As a practical guideline, we summarise attainable classification performance as a function of the event rate and strength of association between the predictors and the response. |
| title | The impact of class imbalance in logistic regression models for low-default portfolios in credit risk |
| topic | Risk Management Computation |
| url | https://arxiv.org/abs/2602.19663 |