Guidelines for Selecting a Forecast Metric

There has always been considerable debate about the best forecast metric to use. All forecast metrics have a place, and there is not a universal perfect fit measure of forecast performance. The author describes 10 guidelines that should be used when considering the most appropriate metric for your organization. |(Actual – Forecast)| Actual 30 Copyright © 2015 Journal of Business Forecasting | All Rights Reserved | Spring 2015 to present aggregate forecast error. Is MAPE (mean of APEs) or WMAPE (weighted mean of APEs) more correct? And of course, there are countless ways to calculate forecast bias. IDIOSYNCRACIES ABOUT THE MEASUREMENT OF FORECAST ERROR Is your head spinning yet? I know mine is. I often attend conferences on forecasting and planning where debates over the correct equation for calculating the measure of forecast error are first formally presented, and then informally argued at luncheons and throughout the cocktail hour. These debates are standard fare, and those of us with many years in the profession have heard them countless times. Every so often, someone will even present a new equation for calculating forecast error that proposes to address all of the other ...