What is the Best Approach?: Evaluating the Quality of Aggregated and Personalized approaches in Credit card Anomaly Detection

Abstract

 Utilization of cashless transactions such as credit card transactions has been increased over the past recent years. This reason forces the researchers to create personalized model for each credit card holder, to identify fraudulent transactions, using various approaches. This paper presents the effectiveness of aggregated and personalized approaches in fraud detection. Data set collected from bank transactions are effectively used to compare the predictions driven by the above said approaches. Naïve Bayes & Random Forest classifiers are effectively used in this research.