Optimization Accuracy of Credit Card Fraud Prediction using Supervised Learning Technique

Abstract

Fraudsters are now more active in their attacks on credit card transactions than ever before. As machine learning and data science have advanced, a number of algorithms have been created to assess if a transaction is fraudulent. Since cloud-based electronic payment methods are used in many internet applications offered by current technology, security and confidentiality are essential. In India, 42% of frauds in a variety of fields were discovered between 1990 and 2020, according to the national herald. Since 1990, the United States' "no fraud" agency has detected about 30% of frauds; each year, these frauds have increased with high ratios. Fraudsters lacked consistent patterns and constantly altered their behavior. Most likely, cloud-based e-commerce and trade business websites are where these scams are identified. To lower this fraud ratio, a genuine and accurate fraud detection system needs to be created. In this investigation, deep learning and artificial intelligence improvement techniques have been applied to identify cloud-based fraud. Although there are numerous works that address this issue, the precision, F-score, review, and precession are remarkably low. Due to this impediment, in this work is introduced machine learning mechanisms like fully Naive Bayes, logistic regression and K-NN. The K-NN is best technique for credit card fraud prediction and achieves good accuracy.