This challenge, contributed by Prof. Robert Fourer, is about assigning people to groups fairly — the same underlying problem as assigning students to project teams or professors to shared offices.
210 people are split across 4 categories (in the original data set: a department code, a work location, a gender, and a job title). They must be assigned to 12 groups, each sized 16, 17, 18, or 19 people, with each person in exactly one group. The goal is “diversity”: prefer assignments where people sharing the same category values end up spread across different groups rather than clustered together.
The interesting question the challenge poses: alongside specialized optimization tools like AMPL, can traditional business rules/decision management tools express and solve a combinatorial problem like this one?
Send your solutions to DecisionManagementCommunity@gmail.com, or open a pull request to add yours here.