Title: A two-stage credit risk scoring method with stacked-generalisation ensemble learning in peer-to-peer lending
Authors: Chongren Wang; Qigang Liu; Shuping Li
Addresses: School of Management Science and Engineering, Digital Economy Research Institute, Shandong University of Finance and Economics, Jinan 250014, China ' SHU-UTS SILC Business School, Shanghai University, Shanghai 201899, China ' School of Management Science and Engineering, Shandong University of Finance and Economics, Jinan 250014, China
Abstract: In this paper we propose a two-stage credit risk scoring approach that can estimate the probability of default (PD) and profit of the loan, stage 1 is credit scoring and stage 2 is profit scoring. In stage 1, the stacked generalisation (stacking) approach is used to train the model. In stage 2, select the loans that are predicted to be non-default in the stage 1, generate a new data set, build a profit prediction model using the stacking algorithm, and introduce internal rate of return (IRR) as the measure of profitability. Finally, the profitability performance of the loan portfolio is studied based on the predicted value of profit. The experimental results show that the predictive performance of the two-stage credit risk modelling approach proposed in this study overcomes the existing methods, and can help investors choose the most profitable loans on the P2P platform.
Keywords: two-stage credit risk scoring; stacked-generalisation; ensemble learning; machine learning; data analysis; internal rate of return; IRR.
International Journal of Embedded Systems, 2022 Vol.15 No.2, pp.158 - 166
Received: 16 Nov 2021
Accepted: 16 Dec 2021
Published online: 08 Jun 2022 *