Forecasting Military Vehicle Spare Parts Requirement using Neural Networks followed by Application of Tacit Knowledge

Abstract

Spare parts forecasting can generally be divided into two approaches that are deterministic and stochastic. Deterministic forecasting is often predictable and often comes from production while stochastic forecasting is often adapted in areas of uncertainties. Even though there are many forecasting techniques available nowadays, the accuracy is often questionable as there may be possibilities of errors especially when it involves inconsistent lumpy demands. The aim of this paper is to look into existing theories with the intent proposal of an alternative method in adapting neural networking where demand patterns from the various external sources would be captured and use to provide an improved prediction method. Therefore, adapting this alternative method would increase the accuracy and confidence level compared to the existing forecasting techniques. This research would also include obtaining tacit knowledge in the purchasing process that is often considered after obtaining the forecasted results prior to the purchase of spares.