[R] plots of ACF

David Winsemius dwinsemius at comcast.net
Fri Nov 21 18:21:59 CET 2008


There are a variety of reasons that a question might go unanswered for  
over a day on R-help. The question may be so technical or narrow that  
only one or two people may be equipped to answer it. Or it may look  
like homework. Or it may lack sufficient detail one which to base an  
answer. Or it may sound like a request for statistical consultation  
more than a request for help with the correct use of R (which is the  
specified purpose of the list.)

I am not a statistician or an r-expert and my experience with time  
series analysis is over 20 years ago, but my impression is that a  
mixture of some of those latter reasons may have been the reason for  
this identical question being unanswered yesterday. You have not  
specified any detail about the units of the serially correlated  
variable. You have not offered any output text. You have not described  
the research question. And there is no real R content.

I can tell you that that time series analysis is well described in  
"Modern Applied Statistics in S" which some people use as their  
beginning R text. If time series was not in your intorductory text  
then maybe you should get one that handles it.

As a complete guess I would speculate that seasonality might explain a  
series with significant correlations at lags of 1, 7, 8, and 9 but who  
knows? You have not given your audience much to work with. And you may  
be outside the boundaries of the list's purpose for existence.

http://www.r-project.org/posting-guide.html
http://www.catb.org/~esr/faqs/smart-questions.html

-- 
Regards;
David Winsemius, MD


On Nov 21, 2008, at 11:45 AM, Sara Mouro wrote:

> Hello.
>
> I have one Model (M3) fitted using the lme package, and I have
> checked the correlation structure of within-group errors using
>
> plot(ACF (M3,maxLag=10),alpha=0.05)
>
> But now I am not sure how to interpret this plot for the empirical
> autocorrelation function.
>
> The problem is that I am used to see/interpret diagrams in which all
> the autocorrelation Lags, except lag-1, are inside the confidence
> envelopes, or those plots where only Lags 1 and 2 are outside those
> envelopes.
>
> But how should I interpret my ACF plot, where Lags 1, 7, 8 and 9, are
> those outside those envelopes?
>
> Is there any correlation structure?
> Which one might that be? AR1()?
>
>
> Also, I have used the plot of the empirical ACF of the normalized
> residuals, but it gives exactly the same results.
>
> Could you please help me?
>
> Best regards
> Sara Mouro
>
>
> Sara Maltez Mouro
>
> Centro de Ecologia Funcional
> Departamento de Botânica
> Universidade de Coimbra
>
> sara at gmesintra.com
> www.uc.pt/ecology/saramaltezmouro
>
>
> Sara Maltez Mouro
>
> Centro de Ecologia Funcional
> Departamento de Botânica
> Universidade de Coimbra
>
> sara at gmesintra.com
> www.uc.pt/ecology/saramaltezmouro
>
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.



More information about the R-help mailing list