[R] Refine ARMA model

[Prof Brian Ripley]

Hint: the 'fixed' argument can be used to set a parameter to a fixed 
value such as zero.

With the reproducible example we asked you for, we might have shown 
you how to use it ....

On Wed, 2 Mar 2011, Chuse chuse wrote:

> Dear users,
>
> I tried to fit an AR(2) model to data. This the result:
>> arima(vw,c(3,0,0))
>
> Call:
> arima(x = vw, order = c(3, 0, 0))
>
> Coefficients:
>         ar1      ar2      ar3  intercept
>      0.1052  -0.0102  -0.1203     0.0099
> s.e.  0.0337   0.0339   0.0338     0.0018
>
> sigma^2 estimated as 0.002934:  log likelihood = 1293.16,  aic = -2576.33
>
> Now, ar2 is not significantly different from zero.
> I would like to refine the model considering ar1 and ar3 only so I fit a model
> x[t]=c+m*x[t-1] + n*x[t-3].
>
> Anyone could help me and tell me how to do it? Thank you very much.
> Chuse

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595