From the course: Business Analytics: Forecasting with Exponential Smoothing

Unlock the full course today

Join today to access over 22,500 courses taught by industry experts or purchase this course individually.

The absolute deviation approach

The absolute deviation approach

From the course: Business Analytics: Forecasting with Exponential Smoothing

Start my 1-month free trial

The absolute deviation approach

“

- [Instructor] When you look for ways to express the degree of accuracy in a forecast, you're looking for a single number that will show how closely all of your forecasts estimate all of your actuals. As a practical matter, a good forecast results in small deviations from the actual observations. So the first solution that might occur to you is to subtract each forecast from its associated actual observation, which results in a deviation. If you do the same with all your forecasts and observations, you wind up with a set of deviations as shown in this worksheet in column E. You could total or average those deviations to come up with the single number that expresses the degree of accuracy in forecasts as show in cell F2. At first blush, the smaller the total or average deviation the more accurate the forecasts. Unfortunately, there's a strong tendency for some of your deviations to be positive numbers and some of your deviations to be negative numbers. It's conceivable that you might…

Contents