A Comprehensive Review of Machine Learning Techniques in Fraud Detection

In the era of technology, most of the peoples are dependent on the latest technology in the form of payments. In the financial sector, the growing number of credit card users presents several important implications, which is leading the users away from the cash payments, the significant increase of credit card use resulted in a decrease in reliance on cash advances. Due to rise in credit card, the financial sector is facing several risks despite having Improved Security and Fraud Detection, Advanced Technology: The adoption of EMV chips, contactless payments, and AI-driven fraud detection systems has enhanced transaction security. These technologies help to protect consumers and merchants from fraud. Detecting financial fraud is complicated by class imbalance, which requires the use of rare data mining approaches along with traditional classification algorithms. To address this, we propose conducting an experimental study to assess the impact of class imbalance and measure the resulting conflict in the imbalanced data. For which we have discussed a variety of papers. These publications, which were collected from sources like Scopus and IEEE Xplore, were chosen using predetermined criteria. These chosen publications were utilized to identify fraud (credit card, UPI, identity theft, fraud loans etc.), with the help of various machine learning methods (KNN, CR7, Gradient boost etc.), the authors' contributions, nations, trends, sources, and datasets used in the tests. The data/reports gathered by different authors used to detect frauds, obtained from the stock exchange and banks of India, China, Canada, the United States etc. One of the foremost contributors of the studies, India, the United States, China, Saudi Arabia remain influential, whereas other countries have a limited number of related publications.