An innovative model-based approach for credit card fraud detection using predictive analysis and logical regression

Abstract

The development of information technology and advancements in communication channels has resulted in increasing fraud throughout the world and immense monetary losses. The objective of fraud detection frameworks is to check each exchange for the likelihood of being false and to recognize fraudulent ones as fast as possible after the fraudster has started to execute a fraudulent transaction, paying little mind to the prevention mechanisms. For this purpose, we utilize a steady foolproof 5 stage verification model with, predictive analysis, logistic regression, outlier model, custom rule management and global profiling. A predictive (LVQ) algorithm alongside logistic regression would improve credit card fraud detection. The benchmark Kaggle dataset is used. The outcomes portray a convincing decrease in credit card frauds.