Description

Ceridian is a global human capital management (HCM) software company that provides work intelligence solutions and services for all kinds of organizations. It uses a modification of Greedy Randomized Adaptive Search Procedure (GRASP) meta-heuristics to solve employee rostering problems. The algorithm is capable of generating feasible schedules, but it has no guarantee of solution quality. Therefore, the client is looking to explore alternative algorithms that can either outperform GRASP in solution quality within a reasonable amount of time or be embedded to help improve GRASP (the baseline).

This project develops a rostering algorithm that is able to give a better solution quality. The output is a roster developed over a 7-day planning horizon, based on employee availability, employee skill sets, appropriate shift lengths, number of consecutive work days, etc. An integer programming (IP) model is formulated to create seven-day rostering schedules. The model aims to minimize the demand coverage penalty (from over-staffing and under-staffing) while enforcing a set of selected fundamental scheduling constraints, such as minimum time between shifts and maximum daily hours.