Forecast models of the form line plus AR noise or cosine plus AR noise with known frequency
fore.sigplusnoise.wge(x,linear=TRUE,method="mle",freq=0,max.p=5,
n.ahead=10,lastn=FALSE,plot=TRUE,alpha=.05,limits=TRUE)
The n.ahead forecasts
The lower limits for the forecasts. zeros are returned if limits were not requested
The upper limits for the forecasts. zeros are returned if limits were not requested
Residuals
The estimated white noise variance based on the residuals
se is the estimated standard error of the k step ahead forecast. zeros are returned if limits were not requested
xi is the kth psi weight associated with the fitted AR model and used to calculate the se above. Note that psi0 is1. zeros are returned if limits were not requested
The variable containing the realization to be analyzed
If TRUE then the program forecasts a line plus noise model. If FALSE the model is cosine plus noise
Estimation method
Frequency of the cosine term. freq is ignored when using line plus noise
Max value of p for the ARp model fit to the noise
The number of steps ahead to forecast
If TRUE then the function forecasts the last n.ahead values of the realization. If FALSE the the forecasts are for n.ahead steps beyond the end of the realization
If TRUE then the forecasts and realization are plotted
Significance level
If TRUE the forecast limits calculated and plotted
Wayne Woodward
"Applied Time Series Analysis with R, 2nd edition" by Woodward, Gray, and Elliott
data(llynx)
llynx.for=fore.sigplusnoise.wge(llynx,linear=FALSE,freq=.1,max.p=5,n.ahead=20)
Run the code above in your browser using DataLab