A generic analytics-driven constructive search heuristic for feasibility in cross-domain timetabling

Timetabling problems are complex combinatorial optimization challenges that arise in various domains, including educational institutions, sports leagues, and healthcare facilities. Despite sharing fundamental characteristics, most solution approaches are domain-specific, requiring substantial redesign when applied to different timetabling variants. We propose a unified ILS-based constructive search framework by extending an Iterated Local Search (ILS) heuristic previously applied to highly constrained sports timetabling. We demonstrate that by solely adapting the domain-specific neighborhood operators to the problem’s structural topology, the core ILS search algorithm can solve diverse timetabling problems while maintaining the same high-level search mechanism, conflict resolution strategies, and parameter values. To validate this cross-domain generalizability, we evaluate the proposed framework on two fundamentally different benchmark datasets: the Toronto examination timetabling benchmark and the International Timetabling Competition 2019 (ITC 2019) course timetabling benchmark. Experimental results demonstrate the consistent performance of the framework, achieving a 100% feasibility rate across all instances in both domains. This performance is comparable to specialized algorithms specifically designed for these respective problems. This work contributes to the development of generic timetabling methodologies, offering theoretical insights into constraint satisfaction across variants and a versatile tool for diverse scheduling contexts. • Propose a generic analytics heuristic for decision-making across multiple timetabling domains. • Demonstrate full feasibility across examination, course, and sports timetabling benchmarks. • Show that one analytics framework supports decision-making with minimal domain adaptation. • Establish cross-domain decision robustness using a unified constructive search method. • Advance decision analytics by reducing reliance on domain-specific timetabling designs.