INTERNATIONAL STUDENTS’ ACADEMIC PERFORMANCE PREDICTION WITH DESCRIPTIVE MACHINE LEARNING EXPLANATIONS

Abstract

Machine learning (ML) models are increasing their penetration in different problem domains
due to their increasing accuracy. Majority of the advanced ML-based stand high in predictive
accuracy but are wanting in terms of explaining their outcomes. This research study has two
objectives. First, to develop predictive models that can predict academic performance of
International students studying in India. Second, to understand the behavior of these ML-based
models in terms of most important features affecting the prediction outcome. The outcomes of
such study are of importance to any university as ability to anticipate performance of these
students can help in planning timely interventions. It is concluded that Attendance of a student
plays an important role towards academic performance. Moreover, understanding of the
behavior of a model help building trust in the model and provide multiple insights of importance
to different stake holders