Detection of Fraudulent Transactions in Credit Card using Machine Learning Algorithms

Today we are living in a digital world where most of the activities performed are online. Fraud transactions are ever growing since the growth of ecommerce applications. Millions of transactions are happening around every second everyday giving us the benefit of enjoying financial services through credit and debit cards. Fraud transactions are allowing illegal users to misuse the money of genuine users causing them financial loss. Accessibility of credit card transactions data, techniques used by the frauds, identifying scams in the bulk data which is getting produced very quickly, imbalanced data are some of the major challenges involved in detecting fraudulent credit card transactions. Hence, we need powerful techniques to identify fraudulent transactions. The current paper deals with various machine learning techniques such as Artificial Neural Network (ANN), Decision Trees, Support Vector Machine (SVM), Logistic Regression and Random Forest to detect fraudulent transactions. Performance analysis of these techniques is done using Accuracy, Precision and False alarm rate metrics. Dataset used to carry out the experiment is taken from Kaggle data repository. The experiment shows that Radom Forest could achieve an accuracy of 99.21%, Decision Tree 98.47%. Logistic Regression 95.55%, SVM 95.16% and ANN 99.92%.