Handling Imbalanced Data for Credit Card Fraud Detection: A Machine Learning Approach

Abstract

Using credit cards has become a routine part of the typical person's life. Credit cards are the most versatile and
convenient method of payment because they are one of the simplest ways to conduct transactions. The process of
identifying credit card theft is both time-consuming and essential. This process should be carried out carefully because
it also affects the interests and flexibility of the customer. Due to the rise in credit card usage and the number of scams
discovered, this study has recently gained significant attention. Despite the high priority of this procedure, it is crucial
to do it efficiently because a big number of customers are engaged. A minor error in the detecting procedure could
result in a significant loss. Predicting the categorical labels of unknown things is the process of classification. In
addition to helping users distinguish between objects of different classes—which are really one single entity on which
operations can be performed—this also helps users organize the data in blocks for simple comprehension. Initially,
the fundamental label categorization is done using the supplied data. Since this classification is just temporary, each
cluster undergoes additional processing to achieve perfection.