An Approach to Detect Credit Card Frauds using Attribute Selection and Ensemble Techniques

Abstract

Managing of an account part is an essential area in our present day era where practically every human needs to manage the bank either physically or on the web Credit-card fraud prompts billions of dollars in misfortunes for online shippers. With the advancement of machine learning calculations, analysts have been finding progressively complex ways to identify extortion, yet handy usage is infrequently detailed. In this paper we are working to identify the fraudulent accounts using classification algorithms and then to improve the accuracy of results using feature selection technique. Bee search and genetic algorithms has been used to select relevant features from large dataset. The reduced dataset has been studied for different aspects. The ensemble learning techniques are implemented to reduce the variance. The impact of bagging, stacking and voting present the optimal technique for fraud detection.