Binary Classification of Students’ Dropout Behaviour in Universities using Machine Learning Algorithms

Higher Education institutions throughout the globe face challenges in maintaining their student retention. Therefore, there is a need of a mechanism to find students who might be susceptible to leave their studies. The research paper focuses on minimising the occurrences of university dropouts in higher educational institutions by identifying students who maybe at a higher chance of failing in the university path. The research employs a dataset acquired from an institute, encompassing information that was accessible at the beginning of students’ enrollment, including their university background, demographic characteristics, and socio-economic circumstances. The dataset consists of Target variable categorized into Graduate, Dropout and Enrolled classes. The important features with respect to Target variable were extracted for building the model. The model is built using the observations of Graduate and Dropout class in the dataset for constructing the model and then classify the Enrolled class students into Graduate and Dropout class. Eight machine learning algorithms were deployed for building the model and best algorithm was selected for classification of the enrolled students. These findings would enable university administration to deploy prompt assistance programmes to aid the currently enrolled students which are likely to dropout in future in order to minimize the number of dropouts in university.