Using Data Science to Prevent Fraudulent Cryptocurrency Transactions over Blockchain Network

Abstract

This dissertation investigates the application of data science, artificial intelligence (AI), and machine learning (ML) techniques to enhance blockchain security, with a focus on preventing fraudulent cryptocurrency transactions. Blockchain technology, despite its transformative potential across sectors, faces critical security vulnerabilities such as 51% attacks, smart contract flaws, and double-spending risks. Addressing these challenges, the research develops an advanced AI/ML-based framework to detect and prevent fraud in blockchain ecosystems, combining pattern analysis, risk scoring, and adaptive learning mechanisms. The study employs a mixed-methods approach, integrating quantitative data from blockchain networks with qualitative insights from case studies. The real-world transaction data from Bitcoin, Ethereum, and Binance Smart Chain networks are analyzed to identify fraudulent patterns and inform the development of AI/ML models. The proposed framework demonstrates a fraud detection accuracy of 97.5% while maintaining an average processing time of 244 milliseconds, outperforming industry benchmarks. The key contributions include a multi-layered risk assessment system, practical testing across blockchain platforms, and the integration of predictive analytics for real-time fraud prevention. The research emphasizes the potential of interdisciplinary approaches that combine data science with traditional cybersecurity practices. These findings offer actionable insights for improving blockchain security, contributing to the reliability and trustworthiness of digital transactions. The future directions focus on scaling the framework for broader applications and integrating advanced AI techniques to address emerging blockchain threats.