Fuzzy Rule-Based Analysis of Spatio-Temporal ATM Usage Data for Fraud Detection and Prevention

Abstract

We present a novel approach for detecting fraudulent behaviors from automated teller machine (ATM) usage data by analyzing geo-behavioral habits of the customers describe the use of a fuzzy rule-based system capable of classifying suspicious and non-suspicious transactions. We first compute the geographic entropies of ATM cardholders to form customer classes based on these entropies. ATM transactions are spatio-temporal by inclusion of location information. The transition data can be generated by using transaction data from the current location to the next one. Once, the transition data are generated, statistical outlier detection techniques can be utilized. On top of classical methods, we can easily use crisp unsupervised methods to detect outliers in the transition data. In addition, fuzzy C-Means algorithm can also easily be implemented to determine outliers. We analyze ATM usage dataset which contains around two years' worth of data, provided by a mid-size Turkish bank. We have shown that a significant bulk of ATM users does not leave the vicinity of their living places. We also present some insightful business rules that can be extracted from geo-tagged ATM transaction data by means of using a fuzzy rule-based system.