A FUZZY RULE BASED INFERENCE SYSTEM FOR EARLY DEBT COLLECTION

Abstract

Nowadays, unpaid invoices and unpaid credits are becoming more and more common. Large amounts of data regarding these debts are collected and stored by debt collection agencies. Early debt collection processes aim at collecting payments from creditors or debtors before the legal procedure starts. In order to be successful and be able to collect maximum debts, collection agencies need to use their human resources efficiently and communicate with the customers via the most convenient channel that leads to minimum costs. However, achieving these goals need processing, analyzing and evaluating customer data and inferring the right actions instantaneously. In this study, fuzzy inference based intelligent systems are used to empower early debt collection processes using the principles of data science. In the paper, an early debt collection system composed of three different Fuzzy Inference Systems (FIS), one for credit debts, one for credit card debts, and one for invoices, is developed. These systems use different inputs such as amount of loan, wealth of debtor, part history of debtor, amount of other debts, active customer since, credit limit, and criticality to determine the output possibility of repaying the debt. This output is later used to determine the most convenient communication channel and communication activity profile.