​YAHNIAN: Bring class registration into the 21st century

The current model is inefficient and frustrating

Data are everywhere. Interpreting those data can be a daunting process. Predictive analytics allow for this vast library of collected data to be repackaged in a more readable manner while describing how to eliminate inefficiencies and optimize a given problem. Andrew Powell, a former University of North Caroline student body president, detailed his vision for a smarter class registration system that blends big data and airline companies. Combining the University’s access to course enrollment, drop, completion and room location data with airline companies’ overbooking practices, we have the opportunity to build a class enrollment system that is more efficient and effective.

Airlines everywhere use data analytics to maximize their productivity. If you’ve travelled on an airplane, then you’ve probably heard a disgruntled gate employee grumble on the loudspeaker that the airline was offering a $500 voucher to take a later flight. This recurring event occurs because airlines regularly overbook their flights in an effort to maximize the number of people on the plane (with deadly accuracy I might add). Since many people book flights multiple months in advance, cancel a few days before or even miss their connection, airline companies use algorithms to predict the number of tickets the airline should sell with the goal of exactly filling the plane’s capacity. When too many people cancel, the airline fills as many of the remaining seats as it can with stand-by seekers. On the flip side, when too few people cancel their flights, the airline compensates denied flight goers with generous incentive packages.

Why is the University any different? We have a class registration system in which many students enroll months in advance, many drop after a few days of class and many don’t even go to class. The simultaneous phenomenon of the virtual bloodbath that ensues to get into a class and the staggering number of empty seats by the middle of the semester must come to an end. The Course Forum is a step in the right direction. The website, used by over 85 percent of University students, presents grade distribution data in an easily readable manner and allows users to make informed decisions. Essentially, I am calling for something very similar. Using all the different course data to which we have access, a user-friendly system would predict the optimal number of students to enroll. In a class that normally caps at 100 students, for instance, by using past data the system would predict how many students will drop and then advise the teacher to over-enroll by X amount of students. The use of a predictive analytics system would save money, make course enrollment more efficient and, most importantly, lead to more students getting into the classes they want. This idea is something the administration, faculty and students can get behind. The best part of this overbooking system is its ability to grow smarter every day: with more data points added each semester, the system would continue to get more accurate in its optimization predictions.

Although this situation seems like a wonderful “win-win-win,” what happens if a class continues to have an over-enrollment when not enough students drop the class? To this problem, the University could employ two strategies. First, the University could over-enroll by two fewer students just to be safe. Secondly, from the financial savings the University will reap, it could divert a small portion to create an incentive package to entice a few students to drop. For these unusually over-enrolled class, the University could offer Cavalier Advantage credit and a guarantee for the class the following semester in exchange for that student dropping the class. While the system isn’t perfect, again it is far superior to our current world without it.

Many teachers deserve praise for already employing this kind of thinking without the need of an analytics tool to help. Using their past experience, some University professors have over-enrolled classes by roughly estimating how many students in the future will drop the class or forgo a lecture. Anyone can go on Lou’s List and watch large Economics or science classes go from an over-enrollment of 30 to the class limit within a week or two. We need, as a University, to champion this kind of efficient thinking to make our class enrollment system smarter.

We have hundreds of talented math, computer science and economics majors prepared and capable of building this analytics tool. I implore you to accept the challenge. The biannual waitlist frustration can be fixed. A higher student satisfaction is within our grasp. All we have to do is ask for it.

Ben Yahnian is a Viewpoint writer.

related stories