An Analysis of the Consumption Demand Data on The DLA's Lead-time Forecast Accuracy

Abstract

The Defense Logistics Agency generates forecasts for demand from military service warehouses (wholesale demand) using a double exponential smoothing (DES) forecasting model. However, the Agency will be moving to a Fourier forecasting model, which explicitly accounts for seasonality in demand data. We found that obtaining additional demand data from the military maintenance centers (consumption demand) will improve lead-time forecast accuracy in both forecasting models. In addition, we found that different incorporations of the consumption demand data are needed depending on the accuracy metric utilized.