mase {bvhar} | R Documentation |
Evaluate the Model Based on MASE (Mean Absolute Scaled Error)
Description
This function computes MASE given prediction result versus evaluation set.
Usage
mase(x, y, ...)
## S3 method for class 'predbvhar'
mase(x, y, ...)
## S3 method for class 'bvharcv'
mase(x, y, ...)
Arguments
x |
Forecasting object |
y |
Test data to be compared. should be the same format with the train data. |
... |
not used |
Details
Let .
Scaled error is defined by
so that the error can be free of the data scale. Then
Here, are the points in the sample, i.e. errors are scaled by the in-sample mean absolute error (
) from the naive random walk forecasting.
Value
MASE vector corresponding to each variable.
References
Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. International Journal of Forecasting, 22(4), 679–688.
[Package bvhar version 2.0.1 Index]