Student Performance Prediction

We propose a data mining method combining feature engineering, gradient boosting, and neural networks to predict student performance. Our prediction achieves an R-squared value of 0.886 on the Kaggle test dataset which is powerful.