Student Performance API with Machine Learning Classification

Design a Laravel-based RESTful API endpoint that ingests student assessment data and uses predictive classification to generate early warning signals for at-risk students. The API should accept JSON payloads containing grade history, attendance records, and engagement metrics. Implement machine learning model integration using PHP's scikit-learn bridge, with endpoints for training, prediction, and model versioning. Include comprehensive error handling for invalid data inputs and provide a detailed response structure that includes risk probability, recommended interventions, and confidence scores.