Fraudulent detection model using machine learning techniques for unstructured supplementary service data

Abstract

The increase in mobile phones accessibility and technological advancement in almost every corner of the world has shaped how banks offer financial service. Such services were extended to low-end customers without a smartphone providing Alternative Banking Channels (ABCs) service, rendering regular financial service same as those on smartphones. One of the services of this ABC’s is Unstructured Supplementary Service Data (USSD), two-way communication between mobile phones and applications, which is used to render financial services all from the bank accounts linked for this USSD service. However, fraudsters have taken advantage of innocent customers and their security vulnerabilities on this channel resulting to high impart of fraud cases, there is still not an implemented fraud detection model to detect these fraud activities. Existing fraud detection models in USSD are in the abstract level and without implementation. Some of the existing studies uses Bayesian’s algorithm for detecting the fraudulent transection. However, Bayesian uses a probabilistic model to predict its output which is influenced by prior history and has a long-term memory which results in low accuracy. This study aims at investigating the design of Fraud detection model using machine learning techniques for Unstructured Supplementary Service Data based on short-term memory. Statistical features were derived by aggregating customers activities were derived using a short window size to improve the model performance using selected machine learning classifiers. To achieve this aim, the research framework consists of two phases, the first phase was data pre-processing and feature derivation. The second phase is model construction and model evaluation. Feature selection was used to select the best set of features for training the model. Many classifiers were trained to investigate their detection accuracy performance. Results of each classifier were tabulated and compared against each other. The results demonstrated that the proposed Fraudulent detection model using machine learning techniques for Unstructured Supplementary Service Data achieve its best performance with Random forest having the best result of 100% across all its performance measure, KNeighbors was second in performance measure having an average of 99% across all its performance measure while Gradient boosting was third in its performance measure, the achieved accuracy is 91.94%, the precession is 86%, the recall is 100% and f1 score is 92.54%. The result validates that with the right feature derived and appropriate machine learning algorithm the proposed model offers the best accuracy in fraud detection. The proposed fraud detection model can help in detection the USSD based frauds for low-end customers who don’t have smartphones.