Measuring the Quality of Intermittent-Demand Forecasts: ItÕs Worse than WeÕve Thought!

In this eye-opening article, Steve Morlidge shows that when our demand histories are intermittent, we should rethink the use of our most common accuracy metrics for selecting a best forecast method. The problem is acute because many software applications use these metrics for performance evaluation and method selection; in doing so, they potentially provide us with poor feedback and inferior models, resulting in harmful consequences for inventory management. Copyright International Institute of Forecasters, 2015