Fraud detection by credit companies is essential in this digital era where the majority of financial transactions are made online. Fraudsters use loopholes in the payment systems to their benefit. Such problems can be solved to a large extent if the companies add an extra layer of security before confirming the transactions using machine learning algorithms. This project intends to use the Isolation Forest algorithm to enhance the security of credit card transactions by predicting the credibility of the transaction before authorization. Detecting 100% of the fraudulent transaction, minimizing the incorrect fraud classifications, and making the process automated is our objective.