University Timetable Scheduling with Constraint Solving

The problem departments have with timetables

University and college timetabling is a hard combinatorial problem, and it is still often solved by one person, a spreadsheet and several weeks before each semester. The result usually contains clashes discovered in the first week of teaching: a teacher in two rooms, a lab section in a lecture hall, a student group with three classes back to back and then a four hour gap. Every late change (a new section, a teacher on leave) means redoing a chunk by hand.

What already exists and the gap

• UniTime (Apereo Foundation) is a comprehensive open source university timetabling system for course and exam timetables, room sharing and student scheduling.
• FET is free timetabling software for schools and universities; Universiti Malaysia Perlis reported using it for course timetabling since 2012.
• Timefold (formerly OptaPlanner) is an open source planning solver with school timetabling examples.
• aSc Timetables is widely used commercial timetabling software, mostly aimed at schools.

These tools are powerful, but they are either large systems that need a dedicated team to configure, or desktop tools built around school structures. When no timetable exists, most give little help understanding why. The gap is a smaller web tool that models how a typical department actually schedules, explains infeasibility in plain words, and respects manual edits.

What you build

Two roles: a timetable administrator who enters data and constraints, runs the solver, locks and adjusts; and teachers and students who view published timetables. Main flows: import courses and rooms from CSV, set availability and preferences, solve, review clashes and scores, lock, re-solve, publish.

The technically interesting part

Writing your own solver would be a different and much harder project. The contribution is modelling:

• Expressing messy requirements as constraints. Labs need lab rooms and consecutive slots, some teachers only teach mornings, combined sections share a room, Friday prayer break is blocked. Model each with CP-SAT boolean variables and interval constraints.
• Hard versus soft. Clashes are mandatory; gaps and preferences are weighted objectives. Show how weights change the result.
• Explaining infeasibility. A solver returning no solution is useless to an administrator. Use assumption literals so the solver can return a small conflicting subset, then translate it into sentences such as "Dr Ahmed's availability, the single chemistry lab and three lab sections on Monday cannot all hold". This is what turns an exercise into a tool, and it is what to demonstrate.

How to evaluate it

• Scale: solve time and soft score for instances of 50, 100, 200 and 400 sections, generated from a realistic distribution.
• Real data: an anonymised timetable from your own department, or a public benchmark from the International Timetabling Competition; compare clashes and gaps with the manual version.
• Infeasibility: create 15 deliberately impossible instances; report how often the explanation names the true cause and how small it is.
• Administrator test: have a real timetable coordinator make five late changes; record time and how much of the timetable changed.

Suggested plan

1. Weeks 1 to 3: interview your department's timetable coordinator; write every rule down.
2. Weeks 4 to 8: data model, CSV import, basic hard constraint model.
3. Weeks 9 to 12: soft preferences, locking, benchmarks on generated instances.
4. Weeks 13 to 18: infeasibility explanation and the web interface.
5. Weeks 19 to 24: real instance, administrator test, write up.

Path to a product

The first customer is your own department, then other departments in the same university. Private colleges and school networks often lack a timetabling tool and pay a yearly licence per campus. A hosted version where the solver runs on your server keeps setup simple, and a consulting service to model a new institution's rules is a realistic early income.

Scope warning

Locking is essential and easy to overlook: a system that discards manual work on every run will not be used. Do not build student registration, attendance or a learning management system. Do not attempt your own metaheuristic solver unless that is explicitly your research question.

Learn what you need on Sefism

• DSA roadmap : graphs, search and complexity behind scheduling.
• DSA cheat sheet : quick reference for the search and graph ideas you will explain.
• MIT 6.042J: Mathematics for Computer Science : logic and combinatorics for modelling constraints.
• Backend roadmap : the API and database around the solver.
• PostgreSQL Documentation : storing constraints, locks and solve runs cleanly.

Get help with this idea

This is a strong advanced project if the modelling comes from a real department. Tauseef can help you decide which constraints to model first, plan the infeasibility work, and design benchmarks examiners will respect. If you are new here, create a free account first.

• Discuss this idea with a mentor
• Send a query to Tauseef
• Book a 1:1 session