Detecting Bank Financial Fraud in South Africa Using a Logistic Model Tree

Abstract

Artificial intelligence is gradually becoming the standard mechanism underpinning online banking. Users' profiles can be confirmed using a variety of methods, including passcodes, fingerprints, acoustics, and images through this technology. On the other hand, traditional cybersecurity measures are unable to prevent internet-based fraud after the visualisation process has been infiltrated. In light of this, the aim of this chapter is to examine the efficiency of the logistic model tree (LMT) in detecting financial fraudulent transactions in South African banks and, ultimately, to develop a financial fraud early warning system. Web-scraping credit and debit card fraud data from SA are used to acquire daily data. The LMT is constructed utilizing a training set from the LogitBoost algorithm and obtained 17 financial conditioning elements. Overall, an early warning system model has shown to be a good performer with a prediction rate of 99.9%. This appears to be a promising approach for detecting online fraud vulnerabilities.