Plant objectives as revealed by shop floor activities
提出一种利用工厂高频运营数据推断管理者对工作调度偏好的新方法，帮助管理科学家将启发式算法更好地应用于实际车间调度，并评估调度方案和预测结构变化的影响。
Abstract One problem management scientists face in adapting heuristics to actual applications on the factory floor is eliciting preferences from plant managers about the importance of different jobs. Even when the goal of the firm is apparently as straightforward as profit maximization, reputation and goodwill can play such critical roles in affecting the revenue flow, that exclusively focusing on a few key variables (such as the price of individual items, their production times, and the raw material cost) may give a distorted picture about which jobs the firm values most highly. This paper offers a new way of eliciting preferences, utilizing high‐frequency data on plant operations that are routinely collected by many firms, in order to infer the direct and indirect cost of scheduling jobs, from actual job schedules that managers reveal by their choices. We then apply our method to a scheduling problem in a steel tube manufacturing plant. After estimating the preferences of the plant manager, we demonstrate how our estimates can be used to evaluate heuristics for hard scheduling problems, and to forecast the effects of structural change, such as expansion in plant capacity, or shifts in job order flow.