What's the right forecasting method? It depends on your data

Matt Leonard, from SupplyChainDive.com wrote a nice news article mentioning some of my recent forecasting research with Fotios Petropoulos (from the University of Bath, UK) and Xun Wang (from Cardiff University, UK). Our article considered the performance of forecasting methods from a supply chain and inventory management perspective. We compared nine different forecasting methods based on the M3 data set, finding that simple forecasting methods… can perform as well as, if not better than, more complex ones. Furthermore, from a supply chain perspective, it is not only forecast accuracy that is important, we should also prefer smooth forecasts that track major trends (without inducing too much bullwhip) whilst filtering out high frequency noise in demand (that should be absorbed in inventory).