Three of the most common measures of forecast accuracy are: mean error, mean square error, and mean absolute percentage error.
Briefly describe the circumstances when you think each might be the most appropriate measure for selecting the 'best' trend curve.
Truly all 3 are tools that should be used to thoroughly evaluate forecasting methods. Each alone doesn't provide validation for a model and so you would never rely on just one of them. However we can think of specific circumstances where you might want to begin with one specific method before applying the others. I will explain the properties of each and provide one example. Try to think of some other examples yourself to make sure you understand the logic.
Mean error is very effective in detecting bias in a forecasting model. If you have a history of forecasts that are always higher than the actual data you will want to use the mean error to quantify the upward bias of the model(s) you have been ...
In a 511 word solution, the definitions of the three terms are carefully explained together with examples.