Labour market areas (LMAs) are a type of functional region (FR) defined on commuting flows and used in many countries to serve as the territorial reference for regional studies and policy making at local levels. Existing methods rely on manual adjustments of the results to ensure high quality, making them difficult to be monitored, hard to apply to different territories, and onerous to produce in terms of required work-hours. We propose an approach to automatise all stages of the delineation procedure and improve the final results, building upon a state-of-the-art stochastic search procedure that ensures optimal allocation of municipalities/counties to LMAs while keeping good global indicators: a pre-processing layer clusters adjoining municipalities with strong commuting flows to constrain the initial search space of the stochastic search, and a multi-criteria heuristic corrects common deficiencies that derive from global maximisation approaches or simple greedy heuristics. It produces high quality LMAs with optimal local characteristics. To demonstrate this methodology and assess the improvement achieved, we apply it to define LMAs in Spain based on the latest commuting data.

劳动力市场区 (LMA) 是一种根据通勤流量定义的功能区 (FR)，在许多国家用作地方层面的区域研究和政策制定的地域参考。现有的方法依赖于手动调整结果来确保高质量，这使得它们难以被监控，难以应用于不同的地区，并且在所需的工时方面难以生产。我们提出了一种方法来自动化划定程序的所有阶段并改进最终结果，建立在最先进的随机搜索程序的基础上，该程序可确保在保持良好的全球指标的同时将市/县优化分配给 LMA：处理层聚类相邻的具有强通勤流量的自治市，以限制随机搜索的初始搜索空间，多标准启发式方法纠正了源自全局最大化方法或简单贪婪启发式方法的常见缺陷。它产生具有最佳局部特征的高质量 LMA。为了演示这种方法并评估所取得的改进，我们根据最新的通勤数据应用它来定义西班牙的 LMA。