Advances in Intermittent Demand Forecasting

Time series of intermittent demand display an erratic pattern that differs from other demand types, with a higher occurrence of zero demands. For this reason, forecasting intermittent demand proves challenging. Most existing methods for forecasting time series of intermittent demand with high intermittency data are unsuitable for generating accurate forecasts. Moreover, we have to take into account the integer nature of data. In particular, given the discrete nature of the data, in this work we suggest employing integer-valued autoregressive models for forecasting. Through empirical evaluation on real dataset, we illustrate significant statistical improvements in the generated forecasts.