A conservative approach for online credit scoring
作者：
Highlights：
• A novel approach for online credit scoring, appropriate for small data and Big Data.
• Implies enhanced non-parametric feature engineering to be more predictive.
• Incorporating a new index increases Statistical and economic performance.
• Inbuilt warning sign for data drift to renew machine learning models automatically.
• Suitable for risky situations, pandemics, and unstable macroeconomic environment.
摘要
•A novel approach for online credit scoring, appropriate for small data and Big Data.•Implies enhanced non-parametric feature engineering to be more predictive.•Incorporating a new index increases Statistical and economic performance.•Inbuilt warning sign for data drift to renew machine learning models automatically.•Suitable for risky situations, pandemics, and unstable macroeconomic environment.
论文关键词：Risk analysis,Online credit scoring,Big Data,Kruskal_Wallis statistic,Open banking,Machine learning
论文评审过程：Received 1 July 2020, Revised 1 March 2021, Accepted 1 March 2021, Available online 10 March 2021, Version of Record 5 April 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.114835
