[R] Extracting AICc and BIC from an ARIMA model.

[David Winsemius]

On Oct 19, 2013, at 10:32 AM, David Winsemius wrote:

> 
> On Oct 19, 2013, at 3:44 AM, Chris89 wrote:
> 
>> Hi everyone!
>> 
>> I am making some time series models, and as i want to compare a lot of
>> models I thought it would be smart to compare the AIC, AICc and BIC values
>> from the models. My question is, how can I extract the BIC and AICc from the
>> model?
>> 
>> As an example:
>> 
>> kings <- scan("http://robjhyndman.com/tsdldata/misc/kings.dat",skip=3)
>> mod = arima(kings, order = c(1,0,0), include.mean=T)
>> 
>> obviously summary(mod) would suffice,
> 
> How would that "obviously" suffice? There is no summary.Arima function, at least that I see. (Maybe you have loaded an additonal package that you were asked to tell us about?)
> 
> 
>> but I don't want to have to look
>> through all the models (as there are alot... )
> 
> The models already have an "aic" element:
> 
>> modList <-list(mod,mod2);  sapply(modList, "[", "aic")
> $aic
> [1] 352.8227
> 
> $aic
> [1] 352.8227
> 
> (Please read the Fine Posting Guide.)

I evemtually did find an hidden and undocumented `summary.Arima` function in the 'forecast' package, but most (all?) of its output was via side-effects. It also appeared that it initially called another hidden function `print.Arima` also using side-effects rather than retruning an object.

library(forecast)
forecast:::print.Arima

The code may be useful if you want to make a parred-down version that spits out AIC, BIC, AICc.


> 
> -- 
> 
> David Winsemius
> Alameda, CA, USA
> 
> ______________________________________________
> 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.

David Winsemius
Alameda, CA, USA