Variable-Work Models for Predicting Course Enrollments

Most higher-education enrollment-forecasting models have been developed to make predictions at the university level or higher. However, this paper discusses some models for forecasting individual course enrollments within departments. They are based on the concept of work; that is, the number of students who have yet to take a given course. We first present a conceptual model that serves as a framework within which our approach can be understood, and then discuss three variations meant to be applied in actual situations. The first two variations were subjected to tests using data from a graduate-level program. Both fared better than various naive alternatives. The more sophisticated of the two demonstrated some improvement over the other, but not in all cases.