This function creates as many "windows" as is possible with the data and calculates an RMSE for each window. The resulting "rolling window RMSE" is the average of the individual RMSEs from each window.
roll.win.rmse.wge(series, horizon = 2, s = 0, d = 0, phi = 0, theta = 0)
The average of the individual RMSEs of each window
The number of windows
The number of observations ahead to be forecasted.
Order of the seasonal difference, default=1
Order of the difference
Vector of AR coefficients
Vector of MA coefficients
Vector of RMSEs ... one for each windwow
The data
The number of observations ahead to be forecasted.
Order of the seasonal difference, default=1
Order of the difference
Vector of AR coefficients
Vector of MA coefficients
Bivin Sadler
"The Time Series Tool Kit"