Paper Contents
Abstract
Online fraud detection is important to maintain financial safety and minimize the risk of transaction. This research uses past transactions data to design a automatic learning fraud system based on random forest regression and logistics regression. By modeling the importance attributes of the transaction, logistics and random forest models determine with high precision if a payment is fraudulent or legitimate. Random Forest offers a strong selection of features along with good unbalanced data management. In this investigation, the random forest classifier and logistics regression were used to identify transactions as fraud or not. The models were tested using primary performance metrics, such as precision, precision, recovery, F1 and ROC-AUC score. From the results, it is clear that the random forest works better than the logistics regression with a precision rate of 98.5% and a ROC-AUC value of 0.998, which shows a better fraud detection capacity.
Copyright
Copyright © 2025 Sayantan Mandal. This is an open access article distributed under the Creative Commons Attribution License.
