Exploring Innovations in Formative Assessment Based on XGBoost

Formative assessment is crucial for real-time student achievement tracking, timely teaching adjustment, and education quality assurance. However, conventional formative assessment is faced with limitations, such as single source evaluator, centralized data vulnerability, and slow feedback. To address these challenges, we propose a secure, transparent, and automated framework with deep ensemble learning algorithm XGBoost as its engine and blockchain backbone as its decentralized and immutable data storage. We implement a four stage workflow, including data collection and on-chain storage, consensus validation, automated scoring by XGBoost, and instant feedback delivery. The experimental evaluation on simulated learning dataset is implemented. Results show that our approach can significantly improve scoring accuracy and drop days of feedback latency to minutes comparing with manual system. With the combination of powerful ensemble learning and immutable ledger, our work provides a practical and theoretical reference for fair, secure, and intelligent formative assessment in modern education.