A hybrid support vector machines and logistic regression approach for forecasting intermittent demand of spare parts

Abstract

Owing to demand characteristics of spare part, demand forecasting for spare parts is especially difficult. Based on the properties of spare part demand, we develop a hybrid forecasting approach, which can synthetically evaluate autocorrelation of demand time series and the relationship of explanatory variables with demand of spare part. In the described approach, support vector machines (SVMs) are adapted to forecast occurrences of nonzero demand of spare part, and a hybrid mechanism for integrating the SVM forecast results and the relationship of occurrence of nonzero demand with explanatory variables is proposed. Using real data sets of 30 kinds of spare parts from a petrochemical enterprise in China, we show that our method produces more accurate forecasts of distribution of lead-time demands of spare parts than do current methods across almost all the lead times.