Automated Student Performance Analytics Pipeline with Machine Learning

Design a comprehensive Python data pipeline using pandas, scikit-learn, and Flask that automatically ingests student grade data from multiple learning management systems, performs predictive analytics on student performance, and generates customizable risk prediction models. The solution should handle disparate data formats (CSV, JSON, SQL), implement automatic feature engineering, create visualizations, and generate automated intervention recommendations for at-risk students. Include robust error handling, logging mechanisms, and a scalable architecture that can process datasets from 1,000 to 100,000 student records.