Auto loan fraud detection using dominance-based rough set approach versus machine learning methods

摘要

Financial fraud is escalating as financial services and operations grow. Despite preventive actions and security measures deployed to mitigate financial fraud, fraudsters are learning and finding new ways to get around fraud prevention systems, thereby, challenging quantitative techniques and predictive models. Thus, new techniques must be explored and tested so the insights obtained from the analysis may be used to support more accurate fraud prediction and the development of fraud prevention systems which have additional checks to mitigate suspicious events. Auto loan is a significant financial product not yet explored in the literature, unlike the misuse of credit cards, for instance. Given the recent increase in fraudulent transactions concerning auto loan applications, this paper tests a new data set for auto loan applications using a technique not yet explored for financial fraud prediction, namely the Dominance-based Rough Set Balanced Rule Ensemble (DRSA-BRE), and after comparing it with other techniques traditionally used for predicting financial fraud, finds that the proposed approach has several advantages over the traditional ones. (C) 2020 Elsevier Ltd. All rights reserved.