Detecting Fraud in Banking Transactions with Random Forest Models
Nataliya I. Boyko, Yaroslav Mokryk · 2021
The methods developed for fraud detection can be applicable for use in modern banking systems that give users the ability to conduct online transactions. With the help of the developed algorithms, such systems would be able to detect fraudulent or anomalous transactions in a timely manner and with great precision, which would allow them to warn users of the system (bank account owners) about the anomalies detected in their transactions and block potentially dangerous transactions with the goal of avoiding unwanted financial losses. The subject of this paper is the analysis and comparison of available methods of banking fraud detection with the methods developed in this paper, as well as the development of the methods themselves and their optimization with the help of various approaches. Also, the subject of research includes the analysis of these optimizations and comparison of them with each other, runtime and numerical analysis of the success of the developed methods and their comparison with the accuracy and speed of methods, approaches and algorithms which had been developed earlier.