Inventory Control of Products at the End of their Lifecycle Based on Nonparametric Methods

Abstract

In this article, we verify that the use of the past stock movement simulation with all combination search, where both control variables are fully discretized, compared to traditional parametric methods, which are often used in management of inventory with sporadic demand, brings economic savings in area of holding and ordering costs. We use sporadic demand data coming from a small size e-commerce company to compare the best achieved holding and ordering costs in continuous review fixed order quantity inventory control policy where the reorder point calculation is based on moving average and linear regression. At the same time, we examine how the results are affected by the required fill rate of service level, which we test for four levels in the interval 25 % - 95 %. The results of our experiments show that AC outperforms traditional parametric methods in achieving the best holding and ordering costs. Moreover, as the level of required service level decreases, the success of AC in achieving the best costs increases. Simultaneously, we see that the success of the simulation increases with increasing variability of demand, i.e. in the case when the differences in quantity between individual non-zero demands increase.