Development of new methods for measuring forecast error

Abstract

Forecast accuracy in supply chain is typically measured using the mean absolute percentage error. This however fails when it comes to measuring the accuracy of forecasts of products having intermittent demand. Mean absolute scaled error (MASE) is a measure of forecast accuracy that works well for intermittent demand and can also be used to compare forecast accuracy between series. In this work, two new methods have been formulated for measuring the accuracy of forecasts with intermittent demand by modifying the MAPE and the SMAPE methods. In this work, the two new methods have been tested with other methods of measuring forecast accuracy such as mean average deviation (MAD), mean average percentage error (MAPE), MASE, mean squared error (MSE) and symmetric mean average percentage error (SMAPE) for datasets of ten products. After performing extensive calculations using datasets from ten products it has been found that the modified SMAPE method performed at par with the MASE method and may be effectively used for measuring the accuracy of forecast of the sales of the various industries with products having intermittent demand.