Forecasting intermittent demand using the cox process

Abstract

If a demand has infrequent demand occurrences and irregular demand sizes, then it is intermittent demand. Generally, intermittent demand appears at random, with many time periods having no demand. Owing to peculiar characteristics of intermittent demand, demand forecasting for intermittent demand is especially difficult. There are ad hoc methods developed for intermittent demand forecasting. Since Cox process has shown superior performance for intermittent demand forecasting, we studied forecasting intermittent demand using Cox process in this study. We develop a new method for estimating Cox process intensity which is called Reversed Leven and Segerstedt (RLS) method. Moreover, we propose a novel method which is a Wavelet Transform and Reversed Leven and Segerstedt conjunction model for intermittent demand forecasting using Cox process. Using real data set of 500 kinds of spare parts from an aviation sector company in Turkey, we show that our method produces more accurate forecasts than other intermittent demand forecasting methods using Cox process. The comparison approach has a lead time perspective which is based on lead time ahead demand forecast and lead time demand forecast errors.