An Efficient Credit Card Fraud Detection Model Based on Machine Learning Methods

Abstract

With the advent of modern technological advances as well as the modern communications expressways, credit card fraud has been rising substantially. Designed to detect fraud of credit card purchases seems to be a major theme to fundamental economic consequences in financial analysis. Credit card fraud tends to cost millions of dollars per year to consumers and also the financial firm. The fraudsters are continuously seeking out new guidelines as well as strategies to commit illegal activities. Therefore fraud protection technologies have now become important to eliminate the losses of banks and other financial institutions. Throughout this research article, we introduce an effective credit card fraud detection mechanism including a feedback system, dependent on machine learning methodology. Its feedback approach contributes to enhancing the classifier's detection rate as well as cost-effectiveness. Afterward examined the performance of different methodologies incorporates random forest, tree classifiers, artificial neural networks, support vector machine, Naïve Baiyes, logistic regression and gradient boosting classifier strategies, on a slightly skewed credit card fraud data sets. These data sets include transaction data through credit card emerges from European account holders with 284,807 trades. Similar approaches apply towards both raw including and pre-processed content. The efficiency of the approaches has always been evaluated depending on just the performance assessment dimensions for different classifiers, which will include precision, recall, F1-score, accuracy, and FPR percentage.

Keywords: Machine learning methods, Credit Card Fraud Detection, Classification method, Supervised learning.