Predicting Achievement of Students in Smart Campus

Abstract

Isolate data among different campus information systems and not much effective information among the big data generated by these systems cause that it is a challenge for predicting achievement of students. This paper designs a student achievement predicting framework, which includes data processing and student achievement predicting. In the data processing, data extraction, data cleaning, and feature extraction are designed. Using these data in data warehouse, we propose a layer-supervised multi-layer perceptron (MLP)-based method to predict the achievement of students. Supervisions are fed to each corresponding hidden layer of MLP to improve the performance of student achievement prediction. Compared with SVM, Naive Bayes, logistic regression, and MLP, our method gets a better performance.