Anomaly Detection in Credit Card Transactions using Machine Learning

Abstract

 Anomaly Detection is a method of  identifying the suspicious occurrence of events and data  items that could create problems for the concerned  authorities. Data anomalies are usually associated with  issues such as security issues, server crashes, bank fraud, building structural flaws, clinical defects, and many more.  Credit card fraud has now become a massive and significant  problem in today's climate of digital money. These  transactions carried out with such elegance as to be similar  to the legitimate one. So, this research paper aims to  develop an automatic, highly efficient classifier for fraud  detection that can identify fraudulent transactions on credit  cards. Researchers have suggested many fraud detection  methods and models, the use of different algorithms to  identify fraud patterns. In this study, we review the Isolation  forest, which is a machine learning technique to train the  system with the help of H2O.ai. The Isolation Forest was  not so much used and explored in the area of anomaly  detection. The overall performance of the version evaluated  primarily based on widely-accepted metrics: precision and  recall. The test data used in our research come from Kaggle.