Spare part demand forecasting in every phase: a data pooling approach to the Bass life cycle model

Abstract

Demand forecasts that capture the life cycle phases of demand are crucial for many high-stake operational decisions. However, difficulty arises when the demand history is restricted to the earlier phases of the product life cycle. An additional challenge occurs if demand is low volume and intermittent, as is common for products in aftermarket industries. In this paper, we present methodology to apply the Bass life cycle model to spare part demand. Furthermore, we propose an extension which pools the incomplete demand history of other products to improve forecast accuracy when a limited amount of demand history has been observed. Our numerical findings show that our extension improves forecast accuracy, even for cases before the peak of demand has been observed. We validate our approach on 175 automotive spare parts and find that pooling the incomplete demand history of multiple products delivers superior forecasting performance.