Design and development of a forecasting module: Case of a warehouse management system

Abstract

We work in this paper on the improvement of the replenishment module implemented on a warehouse management system. This module is based on a forecasting system and a set of inventory management policies. We focus on leveraging the forecasting system to enhance the performance of the software. The current forecasting method used is the simple moving average. This is a well known technique and widespread in practice among logistics managers. However, it is not appropriate for usual consumption patterns including trend and seasonality. We proposed a more advanced forecasting system based on two components: a pattern identification procedure, using statistical tests, and a set of different forecasting models, variants of the exponential smoothing. We made experimentations using historical data from a pharmaceutical platform in order to compare our proposal with the existing one. This enables us to prove the effectiveness of the proposed system in terms of accuracy consumption estimation.