tswge (version 2.1.0)

est.garma.wge: Estimate the parameters of a GARMA model.

Description

This function uses the grid search algorithm discussed in Section 11.5 of Applied Time Series Analysis with R, second edition by Woodward, Gray, and Elliott

Usage

est.garma.wge(x,low.u,low.lambda,high.u,high.lambda,inc.u,inc.lambda,p.max,nback=500)

Value

u

Estimate of u

lambda

Estimate of lambda

phi

Estimates of the pth order AR component of the model where p is some integer from 0 to p.max

vara

The estimated white noise variance

aic

The aic value associated with the final model

Arguments

x

Realization to be analyzed

low.u

The lower limit for u in the grid search

low.lambda

The lower limit for lambda in the grid search

high.u

The upper limit for u in the grid search

high.lambda

The upper limit for lambda in the grid search

inc.u

The increment, e.g. .01, .001, etc. in the grid search on possible u values

inc.lambda

The increment, e.g. .01, .001, etc. in the grid search on possible lambda values

p.max

Maximum value of p allowed for the AR component of the model

nback

Number of backcasts to be used (see section 11.5 in Applied Time Series Analysis with R, second edition by Woodward, Gray, and Elliott

Author

Wayne Woodward

Details

We assume q=0 and do not allow moving average terms in the model.

References

Applied Time Series Analysis with R, second edition by Woodward, Gray, and Elliott. See also Hosking (1984), Gray, Zhang, and Woodward(1989), and Woodward, Cheng, and Gray(1998)

Examples

data(llynx)
est.garma.wge(llynx,low.u=.4,high.u=.9,low.lambda=.2,high.lambda=.4,inc.u=.01,inc.lambda=.1,p.max=1)