Associate Dean, Master's and Bachelor's Programs A MATHEMATICAL PROGRAMMING TECHNIQUE FOR SCHEDULING COURSES AT THE SLOAN SCHOOL

Many schools and universities have been able to schedule resources more efficiently by using formal mathematical models. The course schedules generated by such models may be preferred by all of the school's major constituencies: faculty, students, and the administration. This thesis describes the development of a mathematical programming model used to schedule courses at the Sloan School. The available literature on the general timetabling problem is reviewed to determine the applicability of the various methods to the Sloan School problem. A Decision Support model based on linear programming techniques is developed. That model is distinguished from many of the others in the literature by its completeness. Alternative schedules are generated for the actual Fall, 1987 semester. The best model-generated alternative is shown to be apparently superior to the one generated by the existing manual process. Means of comparison include an improvement in faculty satisfaction and the avoidance of conflicts between courses likely to have a high degree of overlap in student demand. I conclude that the model represents a viable approach to course timetabling at the Sloan School and I discuss recommendations for the next steps required to implement it more formally in the future