Maximising student success with intelligent course recommendations
- Machine learning can recommend courses that students will succeed in
- Intelligent course recommendations make registration easier and keep students on track
- For institutions, this translates to better graduation rates and student satisfaction
The use of machine learning (ML) has been expanding rapidly in the tech world and is primed to effect change in many aspects of the higher education space. Previously, I explained how machine learning could be used to provide more accurate information to university personnel, particularly in admissions, providing accurate enrolment projections. In this article, I’ll explore the next level of automation made possible by machine learning: making intelligent course recommendations.
Currently, intelligent recommendation engines are all around us. When I shop on Amazon, I get recommendations for other products I may like based on my purchase history. While skimming the front page of YouTube, I see video recommendations based on my watch history. In a similar fashion, we can use ML models in the course registration process to make individualised course recommendations to students, helping students choose the courses that best meet their needs, play to their strengths, and keep them on the path to graduation.
Making intelligent course recommendations
Choosing which courses to register for each semester can be complicated for many reasons: the sheer number of courses available is overwhelming, prerequisites are required, or courses are only offered in certain semesters. Additionally, students have to plan their course schedules around other commitments such as extracurricular activities, internships, and jobs. Selecting an optimal course schedule can be time-consuming and complex.
What should the criteria be for selecting which courses are recommended? In the examples of Amazon and YouTube recommendations above, each serves a particular purpose. Amazon’s recommendations seek to maximise the amount of money we spend, while YouTube’s increase the amount of time we spend watching videos.
When recommending courses, we want to optimise for student success. So, we would want to recommend courses that students would succeed in, increasing their confidence and improving their overall experience. And we would want to recommend courses that meet major requirements to keep students on the path to graduation. Helping students graduate on time is the ultimate goal of making intelligent course recommendations.
In order to train an ML model for this task, we provide it with data on many different students and their outcomes in a variety of courses. After training, the ML model would be able to predict how students will perform in particular courses. Then, the student’s unmet major and general education requirements would be assessed in order to make the final course recommendations.
Optimising course schedules
While recommending courses based on graduation requirements and success rate is useful, we would want to also include a scheduling component to suggest classes at times when the student is available. We could ask students to input any potential scheduling conflicts, such as internships or extracurricular commitments, and then only recommend courses that fit within the student’s time constraints.
We would then use an ML model to recognise scheduling features that lead to success for students. For example, some students are better served by morning classes than others. The ML model could look at the student’s performance history to determine the best time of day for that student to take classes. The model could also be used to uncover patterns that inform the schedule, such as whether students who take more than one challenging course on the same day perform as well as students who take them on alternating days. Intelligent scheduling could add another layer of benefit for students.
Enhanced academic advising
It’s important to keep in mind that a course recommendation system should exist to assist and serve students. Students and their academic advisors would still be ultimately responsible for selecting the course schedule that is best for each student. However, the process could be greatly streamlined and simplified by providing a set of course recommendations that students can choose from. We could even recommend several different schedules depending on whether a student wants a lighter academic load that semester or a more ambitious course schedule.
Ultimately, this means that, right from their dorm rooms, students would have assistance navigating through many course options and selecting those that will help them graduate on time. In doing so, students wouldn’t have the burden of extra tuition dollars or the opportunity costs of entering the workforce later. For institutions, this translates to better graduation rates and student satisfaction. Additionally, by using an intelligent course recommendation system, academic advisors could save time and focus on mentoring students in more “big picture” matters.
Now we’ve seen how machine learning can both inform and recommend. In the next and final installment, I’ll explain how machine learning can be used to take action to benefit student retention.