Automated Student Performance Tracking Pipeline with Machine Learning

Design a comprehensive data pipeline that integrates student performance data from multiple Learning Management Systems (LMS) using Apache Airflow. Create an automated workflow that: 1) Aggregates grades from Canvas, Blackboard, and Moodle, 2) Applies machine learning models to predict at-risk students with 85%+ accuracy, 3) Generates personalized intervention reports, and 4) Automatically triggers communication protocols for academic support. Include error handling for data inconsistencies and implement a modular architecture that allows easy integration of new data sources.