Advancing Educational Analytics Using Machine Learning in Romanian Middle School Data

Abstract

This paper aims to present the results achieved after a series of experiments regarding the prediction of the academic performance of Romanian middle school students. A unique data set that was first collected by the authors from 24 pre-university educational units in Romania was used for this study. The dataset contains both rural and urban students, respectively both students with high academic performance and students with low performance. In the experiments, two machine learning methods were used: extreme gradient boosting and support vector machine, along with feature engineering techniques. The obtained results are satisfactory, resulting in an accuracy of 94.18%.