A Hybrid Machine Learning Approach for Credit Card Fraud Detection

1. Introduction

Banking is protecting, screening lens and generate the profits for the customers. Licensed institute to accept deposits to make the large financial services. The services such as growth of wealth, insurance, security and convenience to use of money whenever required. (Terry Turner et al. 2022). The services include investment in home, auto loans together with the discount shopping of the products. A credit card is issued by the financial institute to help the customers to pay for the product in cheapest and the easiest way to support the existing debt along with the borrow of money under some agreement policy. The issuing bank (typically a traditional bank union) establishes a circulating portfolio and provides the user with a credit facility, which even the consumer can use to finance a merchant and then get a cash advance. The plastic chips, stainless steel, gold, palladium titanium and gemstone are the few cards used by the bank.

Stephan et al. 2022 discussed that the finance through credit card is usually defined as the loan from the finance institute within a certain limit which can be access whenever required and whereas required and which can be repaid back immediately or within the certain period of duration with an additional amount of currency in the interest's form. The consumer is verified by the bank with the relation of approval of taking amount. Approval to such intake of currency is done based on the customer’s income.

Amy Fontinelle et al 2021 concluded that the Credit or debit card used for internet shopping has tremendously grown because of the development and positive outlook on the life of E-Commerce, explosion hazard in credit or debit card fraud. Since credit cards are now the most common form of payment with both physical and digital purchases, incidences of credit card fraud are on the rise. In actual situations, prone to fraud coexist alongside cash transactions, and basic pattern recognition approaches are frequently insufficient to effectively reduce fraud. To reduce such losses, all payment card supplying by banks must enforce strict detecting fraud systems.

There are various methods to evaluate the fraud detection in credit card includes technologies of machine learning and deep learning like fuzzy method, genetics and balancing technologies like the smote, Softmax, including different classifier like logistic regression, support vector machine, decision tree and so on.

Figure 1.

Research flow

Figure 1 represents the research flow of the paper.. The collected dataset is the imbalance So, after applying the smote, we have received the dataset in the two form i.e. balance and imbalance. After that, the created dataset is passed through various machine learning classifiers. The same dataset has been applied to the proposed model, comparing the results of all the classifiers. The result contains the precision, recall, accuracy, and F1-score of the evaluation metric to know the better study of it.

This paper contains the following section. Section 2 deals with the studied done before introducing the research work in this field of credit card detection, Section 3 deals with the short introduction of the dataset involved in this research work. Section 4 and 5 deals with the method used and the result of the models. Section 6 gives the conclusion.

2. Related Work

In order to get money or other benefits, fraud is described as an illegal misrepresentation. It is deliberate behaviour that violates the law or even a policy with both the intention of obtaining unfair monetary profit. According to a thorough survey carried out by Clifton Phua and his colleagues, data mining applications and adversarial detection are among the techniques used in this field.

A variety of research is being conducted on new strategies for detecting credit card fraud, and artificial neural network, data analysis, and distributed storage mining are of major interest. To stop this type of credit card fraud, many alternative methods are employed. We can get the conclusion that there are certain additional approaches in machine learning itself throughout detecting credit card fraud after reviewing literature on the subject.

Figure 2.

Approaches for Detecting Fraud Credit Card