Abstract

In recent years, the interest of banks in the relationship with customers has increased, as it is a very important factor for their success, and the challenge facing the bank is the way to retain customers with the most profitability and how to do it at the lowest cost. At the same time, they need to find and implement this solution quickly and more flexibly. The usage of classical methods in data mining in order to detect fraud in various fields, especially, financial. It calls for more complex requirements and takes more time in dealing with different areas of knowledge such as law, economics, business and financial practices. Fraud cases can be similar in content and appearance, but they are usually not identical. The decision tree is used in addition to other methods, which are data mining methods that will be used in the banking sectors and that provide the right product for the right customer with fewer risks. Credit risks such as loan defaults are the main source of risks faced by the banking sector, as data mining methods such as classification and forecasting can be applied to overcome these risks to a large extent or reduce their negative effects. This research presents a forecasting model that serves workers in the banking sector and banks to predict customers who apply for loans from banks and banks. In this research, two models were used to predict the repayment of loans in Iraqi banks, the first is the logistic regression model, which is considered a classic method for forecasting, and the second method is the decision tree method based on the PYTHON program, as the prediction accuracy of the two models was calculated. The tree decision is the best in predicting loan repayments.