In recent decades, new methods and approaches have been developed for forecasting intermittent demand series. However, the majority of research has focused on point forecasting, with little exploration into probabilistic intermittent demand forecasting. This is despite the fact that probabilistic forecasting is crucial for effective decision-making under uncertainty and inventory management. Additionally, most literature on this topic has focused solely on forecasting performance and has overlooked the inventory implications, which are directly relevant to intermittent demand. To address these gaps, this study aims to construct probabilistic forecasting combinations for intermittent demand while considering both forecasting accuracy and inventory control utility in obtaining combinations and evaluating forecasts. Our empirical findings demonstrate that combinations perform better than individual approaches for forecasting intermittent demand, but there is a trade-off between forecasting and inventory performance.

翻译：近几十年来，不连续需求序列的预测方法取得了新的进展。然而，大多数研究仍集中于点预测，对不连续需求的概率预测探索甚少——尽管概率预测对于不确定性下的有效决策和库存管理至关重要。此外，相关文献大多仅关注预测性能，而忽视了与不连续需求直接相关的库存影响。为弥补这些不足，本研究旨在构建不连续需求的概率预测组合方法，在获取组合及评估预测时同时考虑预测精度与库存控制效用。实证结果表明，组合方法在预测不连续需求时优于单一方法，但预测性能与库存绩效之间存在权衡关系。