Automated Student Performance Predictive Analytics Pipeline

Design a comprehensive Python-based machine learning pipeline using pandas and scikit-learn that predicts student performance risk with 85%+ accuracy. The system must integrate historical academic data, attendance records, socioeconomic factors, and learning platform engagement metrics. Implement feature engineering techniques to handle multicollinearity, create a modular architecture supporting real-time prediction updates, and generate interpretable risk scores with confidence intervals. Include a Flask-based dashboard for administrative visualization and early intervention tracking.