Gang Kou: Profit- and risk-driven credit scoring under parameter uncertainty: a multi-objective approach
Reporter：Gang Kou
Abstract：Profit-driven machine learning models and profit-based performance measures have been widely used in credit scoring. When assessing the performance of a machine learning model for credit scoring, previous research typically assumes that the cost and benefit parameters, and their distributional information are available. However, in reality, these parameters and their distributions are often not exactly known. This paper considers the parameter uncertainty in the development of credit scoring models, and the estimation of profits and risks generated by employing those models. We propose a novel profit-based metric—the worst-case expected minimum cost (WEMC)—to estimate the profit of credit scoring models with uncertain parameters. Furthermore, we introduce the worst-case conditional value-at-risk measure (WCVaR) to measure the loss incurred from employing a classification model in credit scoring during the deterioration of cost parameters. A multi-objective feature-selection framework grounded on WEMC and WCVaR is then presented for model development. We employ twelve credit scoring datasets from multiple countries to compare the proposed methods, with feature selection methods that use metrics including the area under the receiver operating characteristic curve, the minimum cost, and the expected minimum cost as selection criteria. The results suggest that the proposed methods outperform other feature-selection methods in terms of cost and risk performance metrics.
