Traditionally, attendance and exam results have been the main way in which educators can show whether or not a student is struggling with the course. This is done retrospectively. With the advent of cloud-based learning technology and online courses the associated metrics are not necessarily the best way to catch at-risk students so that they can be helped.
The converse of that is that this technology can be used to provide and analyse useful data about the students, which can itself highlight those that might be struggling more quickly than can conventional assessment. Moreover, it can do this in a much more timely manner than a retrospective look at attendance and infrequent exam results.
Owen P. Hall Jr. of the Graziadio Business School at Pepperdine University in Malibu, California, USA, describes a machine-learning approach to detecting at-risk students in the International Journal of Social Media and Interactive Learning Environments. “At-risk” is a three-pronged definition alluding to whether a student is considering leaving a course, whether the institution is planning to end the student’s place on the course, or whether they are in a probationary period because of problems they have faced or concerns their teachers have about their course work, attendance, and results.
Machine learning has been used to predict examination grades and even attendance in some educational settings for many years. It is also commonly used to group students for study classes and other activities. It has even been used to detect cheating and plagiarism. It is perhaps therefore not such a great leap to picture the use of machine learning in helping students in another way.
Hall suggests that the machine-learning approach can analyse all the data associated with a student, almost continuously, and determine early on whether a student is at-risk or on the verge of being in that position. At this point, teachers and tutors might intervene to help without delay. The lack of delay to the assistance they give will tend to lead to a better outcome for such students.
“Engaging faculty, educational researchers, and administration in the risk mitigation paradigm is essential for ensuring student success,” writes Hall. Machine learning offers a novel tool to help with this process, improve student outcomes, and reduce dropout rates in an increasingly pressured educational system.
Hall Jr., O.P. (2022) ‘Detecting students at risk using machine learning: applications to business education’, Int. J. Social Media and Interactive Learning Environments, Vol. 6, No. 4, pp.267–289.