A New Decision Tree-Based Model In The Interval Neutrosophic Environment For Detection Fraud In Electronic Banking

The increasing growth and development of banking services in the electronic banking space, as well as the willingness and acceptance of bank customers to use non-face-to-face platforms and portals, have made banks and financial institutions think to expand the platforms needed to provide electronic banking services. . The presentation and introduction of numerous and emerging services of banks in this field, and as a result, the lack of sufficient knowledge of customers to use safe methods and platforms has led to the abuse of fraudsters and brought irreparable losses to customers and banks. Therefore, the main concern of banks is to create security and maintain safe privacy for customers. Also, the accuracy and speed of action in detecting fraud is of particular importance for banks and financial institutions, providing an efficient and effective model while protecting the interests of banks can reduce operating costs and increase the credibility and satisfaction of customers.Generally, fraud detection models follow two types of methods 1- customer behavior history 2- defined frameworks for fraud detection. In this research, in addition to using the usual methods, we tried to detect fraud by introducing a new model in the form of reasoning and mathematical logic. Therefore, to compare the results of the research, we have used a data set on which previous research has been done to detect fraud, which includes the set of online transactions in African banks. In this research, we have processed and analyzed the data by using the principles and rules of interval neutrospheric logic, then we have implemented the proposed model using the widely used CART decision tree algorithm in the Python programming environment to detect fraud.