Assessment of High-Risk Customers to Request a Loan from a Bank with Process Mining

Abstract

This research proposes an approach for identification and assessment of the (suspicious) customers/clients, who are at high risk of taking advantage and getting a loan from a private bank, by means of Disco and Celonis process mining platforms. The identification and assessment process includes the following steps: assessment of the filled application forms, assessment of the submitted documents to the bank, assessment of the possible fraudulent information provided by the customers/clients, revising and improving the quality of the loan offers, proposal of some additional documents which need to be requested to avoid fraudulent requests, and proposal of the content of the documents which need to be revised and changed to prevent clients from taking advantage of them. Accordingly, the rejection and cancellation of a loan request consists of: (1) Starting (2) Scheduling (3) Continuing the process (4) Completing the process (5) Suspending the process (6) Withdrawing from the process (7) Cancellation/Rejection of the request. The findings of the study showed that there have been 303 cases/customers with abnormal/suspicious processes including 14,278 events. It was also found out that, on average, 1 case takes 143 cycles or 24 weeks, to complete the loan request process, while consuming a lot of time and human resources in each activity process. If the results of the study be addressed appropriately, the whole efficiency and effectiveness of the lending process (to distinguish normal clients from the abnormal ones), will be improved, which potentially will result in better and faster identification of fraud.