Student Performance Predictive Model Using Machine Learning

Design a comprehensive Python-based predictive model to forecast student academic performance using historical enrollment data from a large K-12 school district. Develop a machine learning pipeline that integrates student demographics, prior academic records, attendance metrics, and extracurricular participation. The model should predict likelihood of academic success with at least 85% accuracy, generating actionable insights for early intervention strategies. Use scikit-learn for model development, include feature importance analysis, and create a modular script that can be easily integrated into existing student management systems.