Simulation of Stochastic Rolling Horizon Forecast Behavior with Applied Outlier Correction to Increase Forecast Accuracy

A two-stage supply chain is studied in this paper where customers provide demand forecasts to a manufacturer and update these forecasts on a rolling horizon basis. Stochastic forecast errors and a forecast bias, both related to periods before delivery, are modeled. Practical observations show that planning methods implemented in ERP (enterprise resource planning) systems often lead to instabilities in production plans that temporarily increase projected demands. From the manufacturer's point of view, this behavior is observed as an outlier in the demand forecast values. Therefore, two simple outlier correction methods are developed and a simulation study is conducted to evaluate their performance concerning forecast accuracy. In detail, the magnitude of each demand forecast is evaluated and if a certain threshold is reached, the forecast is corrected. The study shows that the application of the outlier correction for forecast values leads to significant forecast accuracy improvement if such planning instabilities occur.