Higher Education
Class clashes per term, down from roughly 833
Average room utilisation, up from 63%
About
The university needed to end timetable chaos across two campuses. Excel grids produced roughly 833 class overlaps a term, room utilisation languished at 63%, and students saw final schedules only three days before classes began. We built a Gurobi-powered integer program that assigns courses, professors, and rooms in minutes, with a Power BI visual planners can drag and re-optimise live.
Industry
Higher Education
Company size
1,000 – 5,000 employees
Founded
2003
The Company
A two-campus university scheduling on Excel grids
The university operates across two campuses with varying room capacities, faculty availability constraints, course prerequisites, and cohort sizes — a scheduling problem with real constraint complexity that the registrar's office was managing entirely in Excel.
Last-mile scheduling decisions — which section goes in which room at which time, with which professor — were made by hand against a spreadsheet grid that had no systematic way to check for conflicts until they had already been built into the schedule.
The challenge
833 clashes a term and a three-day release window
The manual Excel process produced roughly 833 class overlaps per term, and because catching and fixing each one required manual review, the final timetable wasn't published until three days before classes began — leaving students almost no time to resolve conflicts or plan around a late schedule.
Room utilisation averaged just 63%, since the manual process had no way to systematically identify underused capacity across the two campuses, and last-minute fixing consumed significant registrar staff time every term.
The Solution
A Gurobi mixed-integer program with a live drag-and-swap interface
We built a data model spanning courses, sections, cohorts, rooms, time blocks, and professor availability, then defined an objective function eliminating clashes, fitting cohorts to room capacity, and lifting utilisation. Hard constraints (no double-booking, capacity limits, prerequisite sequencing) and soft constraints with penalties (spreading classes through the week, limiting cross-campus travel) feed a Gurobi mixed-integer programming solver with tuned priorities and warm starts.
A Power BI custom visual gives planners a week-grid view where they can drag a section to a new slot and the engine instantly validates and re-optimises, highlighting any resulting conflict rather than letting one slip through unnoticed.
The Results
Zero clashes, 82% utilisation, five-minute schedules
Class clashes fell from roughly 833 a term to zero, and a full conflict-free baseline schedule now generates in under five minutes. Room utilisation rose from 63% to approximately 82%, unlocking latent capacity across both campuses without any new construction.
Scheduling labour fell to roughly 30% of its previous baseline, freeing registrar staff to focus on exceptions and student service rather than manual conflict-hunting, and stabilized timetables meant students received their schedules meaningfully earlier than the previous three-day window.
KEEP READING
Explore more studies on AI, analytics, and enterprise intelligence.
