Comparison of Logistic Regression and Concept Drift Adaptation Algorithms for Modeling Credit Card Data Set with Identification of Accurate Fradulents

Abstract

To predict the accuracy percentage of credit card fraudulent transactions using Logistic Regression algorithm. Materials and methods: Logistic Regression with sample size=100 and Concept Drift Adaptation (CDA) with sample size=100 was iterated different times for predicting accuracy percentage of credit card fraudulent transactions. The sigmoid function used in logistic regression maps the value between 0 and 1 which helps to improve the prediction of accuracy percentage. Results and discussion: Logistic Regression has significantly better accuracy (98.2%) compared to Concept Drift Adaptation accuracy (78.4). There was a statistical significance difference between Logistic Regression and CDA with (p=0.001) (P<0.005). Conclusion: Logistic Regression algorithm helps in predicting with better accuracy percentage of credit card fraudulent transactions.