Detecting students at risk using machine learning: applications to business education

Abstract

Detecting students at risk continues to challenge the management education community. Traditionally, student examination performance and attendance have been two of the primary metrics used for identifying students at risk. However, waiting until midterm exam results to intervene can often prove problematic. With the advent of cloud-based learning platforms, these traditional factors can now be complemented by a variety of quantitative and qualitative metrics. The results from the current study indicate that machine learning-based classification models can detect struggling students and identify appropriate intervention initiatives. Specifically, student performance on practice quizzes was found to be an effective early warning indicator, which, in conjunction with related student attributes, can be used to identify customised amelioration strategies. The primary purpose of this article is to highlight how machine learning can reduce student dropout rates and improve overall learning outcomes throughout the business education universe.