Abstract:

The allocation of teaching assignments in large-scale university programmes is a complex decision-making task that requires balancing institutional priorities, instructor expertise, workload equity and staff preferences. This paper presents a methodical, data-informed optimisation approach developed for the National University of Singapore (NUS) Data Literacy Programme (DLP), a university-wide initiative comprising a suite of data literacy courses for its Executive & Administrative (E&A) workforce.

Within the DLP, the foundational course is mandatory for its E&A workforce while intermediate and advanced courses are optional, with more specialised content and varying demands. These distinctive variations introduce unique assignment challenges across course levels, particularly in ensuring workload balance and alignment between instructor strengths and teaching preferences. The current DLP delivery model is based primarily on paired teaching, which differs from the predominately solo-teaching practices common in many higher education contexts. While pedagogically beneficial, this model substantially increases assignment complexity by introducing collaborative preferences as additional constraints.

To address these challenges, we propose a three-step assignment workflow that integrates managerial judgment with mathematical optimisation. The proposed optimisation framework also enables systematic exploration of partial solo-teaching arrangements for selected classes. Our model incorporates multiple parameters: instructor rank and eligibility, number of classes per course, past teaching feedback scores, pairing preferences, course teaching preferences, instructor workloads and course credit weightings.

Firstly, instructors are manually pre-assigned to selected courses based on strong teaching performance scores. In the second stage, the remaining non-foundational courses are allocated using a mathematical optimisation model that maximises pairing and course teaching preferences while satisfying eligibility and workload constraints. In the final stage, foundational classes are assigned through a separate optimisation algorithm to optimise pairing preferences where equitable workload distribution is treated as the primary constraint.

Results indicate that the multi-stage design improves workload equity, alignment between instructor strengths and course assignments and transparency of decision-making. The staged approach also enhances administrative interpretability and facilitates stakeholder acceptance.

This study contributes a practical and extensible model for academic workforce teaching assignment planning, demonstrating how optimisation techniques can be meaningfully integrated into educational management. The proposed approach is adaptable to other higher education contexts where foundational and elective professional development courses coexist to support more sustainable and data informed teaching deployment practices.

Keywords:

Optimisation modelling, higher education management, academic workforce planning, data-informed decision-making, data literacy programme, adult learning.