Detecting Fraud on the Ethereum Blockchain Using the XGBoost Algorithm

Abstract

In recent years, blockchain technologies such as Ethereum have secured widespread adoption, yet they have also become increasingly targeted by fraudulent activities. Discovering these fraudulent patterns is challenging due to the complexity and size of transaction data. This study investigates the application of the XGBoost algorithm, a gradient boosting technique optimized for performance plus scalability, in discovering fraudulent transactions on the Ethereum network. The model is instructed to distinguish between valid as well as suspicious behaviour based on transactional and behavioural characteristics by examining a dataset of Ethereum transaction records. XGBoost is a popular machine learning algorithm used for a range of tasks, including fraud detection on Ethereum and other blockchain networks. It is a highly effective model due to its performance, flexibility, and ability to handle complex, imbalanced, as well as large datasets. This study emphasises accuracy and precision, also recalling key metrics, demonstrating that XGBoost not only enhances prediction performance but also minimizes false positives over other classification algorithms. Our findings: XGBoost is the optimal model for real-time fraudulent detection in a blockchain context. The accuracy of the XGBOOST algorithm achieves a high level of accuracy.