Search This Blog

Oct 26, 2007

Just-in-time scheduling

We live in the midst of an incredible information technology revolution. We ask questions that were never asked before, we can know things that otherwise simply could not be known. For example, here is a simple question: how many students will need class X in the next semester, and what time of the day will accommodate the most? To answer this question, we usually use the last year’s experience, some informal feedback, and whatever other hunches we might get. To prepare for especially hard-to-plan course like student teaching, we use some sort of application process. However, we can never account for fluctuation in numbers and scheduling conflicts.

Let’s assume the Math school decided to move their MATH 283 two hours down, simply because they cannot have faculty who can teach it during their regular hours. This means that Elementary kids cannot take our EDRD 419 class. So, a lot of them decide to try it in the following semester, and we cancel a low-enrolled class. Yet in the following semester, there is a bubble we have no idea about: kids whose time is it to take 419, plus all those who delayed last time because of Math, all want to take the class. They try to get in, they cannot, and then they start complaining to the Dean. So we realize there is a problem, create a sign-up list, struggle to find an instructor, and finally offer it anyway. All of this is OK, but costly: students are upset, our FT faculty maybe underemployed when we cancel, but then we have to pay an adjunct extra. School Director’s valuable time is wasted.

What we need is a data management system. Students will develop their four-year tentative plans, so the system will know how many students need what when, for at least a couple of years in the future. The closer it gets, the more accurate picture of student demand we will have. We would also have a better idea of our expenses in the future, and could plan our budgets accordingly.

Then if Math 283 will happen to get scheduled first, the system will know that students who need it also need EDRD 419, and will suggest the best times for it and other yet unscheduled classes, so most students do not have conflicts in their schedule. Of course, if we schedule first, Math folks have to use the free time available.

Of course, something like this does exist already; just check out all these products. Yet it does not appear any of them have the capacity to look several years ahead; they basically play one semester in advance, and help match people’s preferences (they ask students and faculty when they would like to teach or take classes). Of course, no student wants to drag his or her behind on campus at 7:30 in the morning, and faculty may have their own preferences. So, the system can keep track of three factors: how many people need a class, when they are available to take it, and when they would rather take it.

It is not really that complex, and probably not that expensive to develop. Other industries such as shipping may have used similar algorithms to manage different processes. Any takers?

No comments:

Post a Comment