Benchmark Dataset for the Multi-Layer Anaesthetist Rostering Problem

This dataset provides a comprehensive benchmark for the Anaesthetist Rostering Problem (ARP), a multi-layer healthcare workforce scheduling problem involving monthly on-call, weekly daytime, and operating theatre room
assignments. The dataset includes 32 synthetic instances across four sizes (Small, Medium, Large, XLarge) and four difficulty levels (Easy, Medium, Hard, VeryHard), along with 5 months of anonymised real-world operational
data from a Malaysian teaching hospital. The constraint model comprises 8 hard constraints and 21 soft constraints with dual-matrix fairness enforcement, all verified against real-world data with zero hard constraint
violations. Baseline results from IBM CPLEX 22.1 and Google OR-Tools CP-SAT 9.12 are provided for all instances, together with an evaluated manual roster for comparison. Source code for data loading, instance generation,
solver implementation, and manual roster evaluation is included to support reproducibility.

本数据集为麻醉科排班问题（Anaesthetist Rostering Problem, ARP）提供了一套全面的基准测试集。麻醉科排班问题是一类多层级医疗人力调度问题，涉及月度待命排班、周度日间排班以及手术室分配任务。该数据集包含32个合成测试实例，涵盖四种规模（小型、中型、大型、超大型）与四种难度等级（简单、中等、困难、极困难），同时附带马来西亚某教学医院5个月的匿名化真实运营数据。其约束模型包含8条硬约束与21条软约束，并采用双矩阵公平性校验机制，所有约束均经过真实运营数据验证，未出现任何硬约束违反情况。针对所有测试实例，本数据集提供了IBM CPLEX 22.1与Google OR-Tools CP-SAT 9.12的基线求解结果，同时附带一份经评估的手动排班表以供对比。数据集还包含数据加载、实例生成、求解器实现以及手动排班表评估的源代码，以保障研究可复现性。