Applying the zero-inflated Poisson regression in the inventory management of irregular demand items

Managers often need to face the problem to define the inventory level of items having an irregular demand. Difficulties arise when the existing methods cannot be applied due to some hypotheses, leading to wrong inventory levels, and, higher holding costs. In this paper, a methodological framework, based on the Zero-Inflated Poisson regression, and called Spare Part Oriented ZIP Approach (SPOZA) is developed. It aims to simplify the inventory-level evaluation procedure when items have an irregular demand pattern The SPOZA output is the evaluation of the optimal theoretical protection level to fulfil to reduce inventory total costs as a percentage value. Aiming to evaluate the benefits of applying SPOZA, a real-case application is presented. Several scenarios are analyzed, and comparisons with standard forecasting methods are conducted. Results are very promising since SPOZA leads to an inventory costs reduction and performs better than other approaches, especially when the items demand increases.