University class-professor scheduling model for the Portland State University computer science department

Summary form only given. This paper presents the University Class Scheduling Model specifically designed for the Computer Science department at Portland State University (USA). The approach taken with this paper uses genetic algorithms employing Ward System´s GeneHunter with Excel for the PC as the target platform. Excel was selected so that non-programmers could easily modify and expand the model as the needs of the curriculum changes. The paper introduces the genetic model developed to automate this process. Once the theory behind the model´s framework is presented, the paper then steps through each of the inputs, results, and constraints required to develop recommended and feasible schedules. This includes constraints for matching faculty to courses, for selecting the day and time a course is offered, for blocking faculty availability so they will not be assigned more than once for a given day and time, and for offering sections specifically in the evening. Each constraint is described in terms of its functionality and is also formulated mathematically. A total of over 600 hours of test suites were executed and evaluated to determine the model´s viability