Student Performance Predictive Analytics Pipeline

Develop a comprehensive predictive analytics pipeline using TensorFlow and pandas that forecasts student academic performance with machine learning models. The system should integrate historical student data, including demographic information, previous grades, attendance records, and extracurricular activities. Implement multiple predictive models (logistic regression, random forest, neural networks) and create a robust evaluation framework with cross-validation and performance metrics.