Metric¶
https://otexts.com/fpp2/accuracy.html
(MAE) mean absolute error
(RMSE) root mean squared error
(MAPE) mean absolute percentage error: not good when value is close to zero
(MAAPE) mean arctangent absolute percentage error : not consistent with the other metrics, why???
For a combination of parameters, if a metric has different patterns compared to the other metrics, it's not consistent and should not be used. For example, all other metrics show a min for a specific combination of parameters but that metric does not show a min.