Credit Card Fraud Detection on Original European Credit Card Holder Dataset Using Ensemble Machine Learning Technique

More and more of our financial transactions are becoming digitalized. Credit cards are therefore the modern default medium for modern financial fraud attacks. Since many financial losses have been suffered due to credit card fraud, researchers working for both public and private institutions have decided to apply artificial intelligence to the problem. Modern attempts at using AI based credit card fraud detection involve using sampling techniques featured in machine learning. Sampling essentially means that a machine learning algorithm learns from a past dataset and tries to extrapolate results from past data sets to present data sets. While sampling can be a useful technique in some scenarios, researchers found that using sampling based machine learning methods often overfit or underfit datapoints and therefore have credit card detection systems of limited accuracy. To address this issue, in this paper researchers decided to use “fundamental machine learning models” instead of sampling based schemes to detect credit card fraud. These original models greatly simplified the process of detecting credit card fraud.