Forecasting Lumpy Demand for Planning Inventory: The Case of Community Hospitals in Thailand

The pattern of lumpy demand affects the healthcare industry such as small community hospitals which encounter sporadic demands of slow-moving medicines. The objective of this work aimed to study performance of forecasting methods for inventory planning. This work compared two forecasting methods: Croston(CR) and the Teunter, Syntetos, and Babai's (TSB) methods. Furthermore, this work proposed a combination of Exponential and Poisson distribution and the use of average inter-demand interval combining with average demand to forecast lumpy demand. It assumed that amount of on-hand inventory would be equal to forecasting values. Mean square error (MSE) and number of shortage period were used as performance indicators of these methods. The simulation used 2 types of vital medicines, obtained from the community hospital in Mae Hong Son province from January 2015 to December 2017. The results from CR and TSB methods provided lower MSEs when the smoothing constants were unchanged until 12 weeks. Meanwhile, adjusting the smoothing constants every 4 weeks provided lower shortages. Meanwhile, the other two proposed methods led to lower shortages comparing with those of CR and TSB methods.