Another look at measures of forecast accuracy for intermittent demand

 

Some of the proposed measures of forecast accuracy for intermittent demand can give infinite or undefined values. This makes them unsuitable for general use. I summarize the various measures and demonstrate what can go wrong. Then I describe a new measure (the mean absolute scaled error) which does not have these flaws. I believe it should become the standard measure for comparing forecast accuracy for multiple intermittent-demand series.

Errata: In the series shown in Table 2, the sixth value should be 11 (not 1). Similarly, the fifth error value should be 11 (not 1).

Sample calculations: Excel spreadsheet showing MASE calculation for the example series.

Data: Data used in examples.