A comparative study of credit card fraud detection using machine learning

Abstract

Credit card fraud is one of the worldwide problems which affects everyone. Credit card fraud detection is a distribution issue with the aim of automatically and adaptively categorizing genuine and fraudulent transactions. Any malicious behavior causing financial loss to the other party is classifying as fraud. For example, in poor nations, the use of digital currency, or even plastic money, is on the rise it has a history of defrauding people. They have a track record of scamming others. In recent years, credit card fraud has increased. Customers and institutions all over the world are paying billions of dollars. Fraudsters continue to thrive despite the multiple fraud-prevention devices in place. In this study report, we're seeking to come up with new ways to swindle people. As a result, combatting these scams demands the implementation of a sophisticated fraud detection system. Fraud is not only detected, but also prevented by the system. The systems must be able to learn from past fraud schemes and adapt to new ones. The notion of credit card fraud, as well as the many types of fraud, were discussed in this study. Several fraud detection methods, such as logistic regression, decision trees, and random forests, were studied. Existing and proposed theories for credit card fraud are carefully scrutinized, and these strategies are tested using quantitative metrics such as accuracy and discovery rate. The system showed a high level of fraud detection, equal classification, a high Matthews correlation coefficient, and a false alarm level. The crime of stealing sensitive information, supplanting, grazing or stealing data on the part of the merchant, lost or stolen cards, producing fake or counterfeit cards, making a real site, and removing or replacing a magnetic line on the card keeping user information are all examples of credit card fraud. The study came to the conclusion that existing models have flaws and proposes a new technique for fixing them. Obstacles to fraud detection are expected to change and metamorphosize into hidden impediments in the future, based on how fraudsters do these illicit behaviors.