Early Detection of Poor Academic Performers Using Machine Learning Predictive Modeling

Abstract

The student's academic development, retention, and attainment gap are considered as the common key factors that influence the institutional academic performance. In this regard, educational institutions are focusing to reduce the attainment gap between good, average, and poor performing students. Two different datasets are taken for this study. Students' data is collected through questionnaire, and Dataset 1 (D1) is created. The second dataset (D2) is taken from the repository. Both the datasets have been preprocessed followed by attribute selection and predictive modeling. In this study, predictive models have been built, and the learners are classified as high, average, and low performers based on their academic scores as well as on their demographic characters. The three classifier models are applied on the datasets. Based on the evaluation measures, the best classifier is identified. This early identification of low performance students will help the educators as well as the learners to put a special care to enhance the learning process as well as to improve the academic performance.