A benchmark dataset for the Anaesthetist Rostering Problem

A comprehensive benchmark dataset for the Anaesthetist Rostering Problem (ARP), comprising 30 synthetic instances across three complexity levels.<br>DATASET FEATURES:• 30 instances (10 small, 10 medium, 10 large)• 84-day planning horizon (3 months × 28 days)• 17 locations (7 monthly on-call + 10 weekly office/evening)• 14 hard constraints + 14 soft constraints• Bi-level optimization structure (monthly → weekly rostering)• Standardized CSV format<br>STAFF COMPOSITION:• Small: 12 permanent staff (8 Junior + 4 Senior)• Medium: 23 permanent staff (14 Junior + 9 Senior)• Large: 41 permanent staff (23 Junior + 18 Senior)<br>CRITICAL UPDATE:This version implements feasibility-aware instance generation to guarantee solvability:• Minimum CCT (Consultant Cardiothoracic) qualified staff: ≥ 3• Minimum SCT (Specialist Cardiothoracic) qualified staff: ≥ 2• 100% Stage 1 feasibility verified with IBM CPLEX 22.1.1 and Google OR-Tools CP-SAT<br>Previous versions (v1, v2) contained instances where insufficient cardiothoracic-qualified staff could cause solver infeasibility. The current version resolves this limitation.