Impact of Postgraduate Students Dropout and Delay in University: Analysis Using Machine Learning Algorithms

Abstract

Cost of education and economic background are some factors that influence student dropout from postgraduate studies. However, high dropouts do not affect the students only, but also impact university revenue. This research analyzes various literature on machine learning algorithms and applies suitable algorithm to produce a prediction model. This study indicates that decision tree and Random Forest algorithms have better accuracy, class recall, and class precision than Naïve Bayes. Therefore, the prediction model uses the Decision Tree algorithm to provide various approaches to maximize revenue in universities. The findings indicate high dropout rates negatively impact university revenue, while low rates influence revenue positively. Other aspects like grants received by students, the number of research publications, anddegree level also positively or negatively impact revenue if the dropout rate is medium. A complete understanding of this prediction model can identify and minimize the risk of early withdrawal or delayed graduation and improve revenue generation by universities.