Detection of Credit Card Fraudulent Transactions Utilizing Machine Learning Algorithms

Abstract

Over the last several years, fraudulent credit card transactions have increased significantly. For financial institutions, credit card fraud is a significant problem, and reliable detection of fraud is generally challenging. According to annual study conducted in 2021, more than 50% of Americans have had experienced credit and debit card frauds, and beyond 13% of individual who utilize these cards do so on a regular basis. This corresponds to 127 million Americans that have roughly once been the victim of credit card fraud. Using the traditional technique, it is highly challenging and time-consuming to detect such fraud occurring across a large database. An effective technique to handle this kind of issue is by using technology such as AI and machine learning to create a fraud detection system which is well automated to identify and categorize such occurrences. This paper presents six supervised machine learning algorithms as Naive Bayes, SVM, Random Forest, KNN, Logistic Regression and XGBoost which is used create a classification model that correctly identifies such fraudulent. After evaluating all these algorithms, finding shows that Support vector Machine is the most reliable model among all in terms of classifying fraudulent and non-fraudulent transactions gaining the highest accuracy results.