Online Transaction Fraud Detection Using Machine Learning

Aditya Sunil Raje, Dr. Ayesha Siddiqui

Abstract: Increasingly widespread use of online payment systems, which include e-commerce websites, Internet banking, mobile transitions, has resulted in a growing number of fraudulent cases and posed great security risks not only to the users but also to financial companies. Traditional methods used for detection of fraudulent activities based on rule set are often insufficient due to their rigidness and no adaptability. In order to solve this problem, this paper offers a machine learning solution to detect fraudulent transactions in real time. Specifically, a number of supervised learning models, such as Logistic Regression, Decision Tree, and Random Forest will be utilized for analysis of transactional behaviour and differentiation between regular and fraudulent transactions. The dataset for this analysis is based on historical transaction data and includes stages of preprocessing, such as cleaning, normalizing and feature selection. The efficiency of the system will be evaluated through a series of characteristics, including Accuracy, Precision, Recall and F1-score. The results indicate the advantage of Ensemble learning models, especially Random Forest method, over traditional rule-based systems in terms of accuracy and ability to handle imbalanced data. Moreover, the model proposed in this paper is able to automatically adapt to newly received transactional data.

Keywords: Online Transaction, Online Transactions, Anomaly Detection, Supervised Learning, Real time Detection, Data Preprocessing, Classification, Financial Security