Machine Learning-Based Credit Card Fraud Detection: Evaluating Standard and Hybrid Techniques with Majority Voting
Abstract
: Financial businesses face a major challenge with credit card theft, which results in billions of dollars in annual losses. Due to confidentiality concerns, there are limited research studies on the analysis of real-world credit card data. This study employs machine learning algorithms to identify credit card fraud. Standard models are used initially, followed by hybrid techniques using AdaBoost and majority voting techniques. The effectiveness of the model is evaluated using a publicly accessible credit card dataset, as well as a batch of actual credit card data from a financial institution. Additionally, noise is introduced to the data samples to assess the robustness of the algorithms. The experimental findings suggest that the majority voting approach performs well.
Keywords
Financial businesses, Credit Card Theft, Fraud Detection, Neural Networks, Machine Learning, Adaboost, Majority voting techniques.