Suitability of Box-Jenkins modeling for Navy repair parts

Abstract

A basic function in the proper management of repair part inventories is the forecasting of future demand. The Navy maintains a database of univariate demand data for its repair part inventories using a quarterly time interval. Historically, Navy repair part demand forecasting has been done using the exponential smoothing procedure. This method is a simple and robust means of forecasting, however it does not make use of any characteristics of the entire time series such as trend, cycles, presence of outliers, or demand clustering. This research begins by developing several simple, robust, and dimensionless time series features. These features are used to predict the suitability of Box- Jenkins (ARIMA) modeling. The ARIMA process is a powerful time series modeling and forecasting technique which possesses flexibility for the inclusion of many time series characteristics. This research project develops a predictive model of ARIMA suitability using both classical regression and a modem expert-system statistical package, ModelQuest. A computationally simple means is presented for determining which time series may benefit from the Box4enkins methodology. Using ARIMA modeling for time series that show significant benefit will provide a more accurate demand forecast and benefit inventory management.