Intelligent Financial Fraud Detection Using Machine Learning Techniques

Abstract

The adoption of online banking in India has shown a sharp increase in online financial transactions, and it has also created new prospects of fraudulent activities. Financial systems need to identify the cases of fraud within a short time and with high precision to protect their structure and safeguard the assets of the customers. In this study, the authors created a powerful machine learning model with the Extreme Gradient Boosting (XGBoost) algorithm to identify banking fraud using the Bank_Transaction_Fraud_Detection dataset of Kaggle. The model proved to be remarkably proficient in the differentiation of legitimate transactions and suspicious transactions with a highest accuracy of 97.2%. The results reveal that XGBoost is a powerful, robust and scalable solution that can be applied in real time detection of fraud in Indian financial institutions. The algorithm analyses the financial records of the dataset to conclude whether the transactions are valid or fraudulent transactions. The model analysis shows that it is capable of identifying fraudulent transactions and at the same time it is capable of differentiating the normal and suspicious financial transactions. The results show that XGBoost is a powerful and scalable system that is applied in the detection of fraud in the Indian financial institutions.

Keywords