Credit card fraud detection by implementing machine learning techniques

Abstract

Application of machine learning algorithms in the fraud detection area for credit card operations has now become an important constituent of research in the domain of digital transactions. The evolution of various machine learning techniques like classification and clustering has shown the requirement for application of related algorithms in detecting frauds of credit card transactions. Real-time detection of fraudulent activities in the financial transactions can help to save huge financial losses as well as the goodwill of the financial institutions to its customers. In this study, the application of various classification techniques has been proposed by using machine learning algorithms for detecting the accuracy of the fraud detection. Some commonly studied classification methods are implemented by considering a large volume of data. Different algorithms that have been evaluated empirically are naïve Bayes classifier, extreme learning machine, K-nearest neighbor (KNN), multilayer perceptron (MLP), and support vector machine (SVM). A machine learning classification model has been proposed by hybridizing SVM, KNN, and MLP classification models where it is observed that the predicted value of accuracy has improved significantly.