Guidelines for choosing a metric

Throughout this chapter, we have come to understand that it is difficult to choose one forecast metric and apply it universally. There are advantages and disadvantages for each metric and being cognizant of these while selecting a metric is the only rational way to go about it.

Let’s summarize and note a few points we have seen through different experiments in the chapter:

• Absolute error and squared error are both symmetric losses and are unbiased from the under- or over-forecasting perspective.
• Squared error does have a tendency to magnify the outlying error because of the square term in it. Therefore, if we use a squared-error-based metric, we will be penalizing high errors much more than small errors.
• RMSE is generally preferred over MSE because RMSE is on the same scale as the original input and therefore is a bit more interpretable.
• Percent error and symmetric error are not symmetric in the complete sense and favor...