“Credit Card Fraud” Detection Using Data Analytics A Comparative Analysis

Abstract

Fraud events take place frequently which results in a huge financial loss. Fraud detections are dynamic and are not easy to identity. Data mining plays a vital role in detection of "Credit card fraud" done in fraudulent online transactions. Fraudsters use latest advanced methods which is an advantage. This process becomes challenging based on two major reasons -firstly, the profiles of users keep changing constantly and secondly, the datasets required for this are highly confusing. The overall performance of "Credit card fraud" detections is improved by sampling approach on the dataset. This research looks at fraud incidents in the context of real-life fraud transactions. A variety of machine learning methods and modelling approaches are used to combat each fraud. The goal of this study is to see how well logistic regression and K-nearest neighbour (KNN) perform on highly skewed "Credit card fraud" data. In order to assess the algorithm's robustness even further, noise is injected into the data sets. The major purpose of this study is to compare and contrast numerous methods for identifying fraud.