Re: Time series prediction

[Freddy López]

Hello miro99_ale,

It is possible your data were generated from any general processes than (stationary) ARMA(X) but this is not the subject here... :) 

Now, if you are looking for a description for those, please use ACF and PACF plots and inspect numerical autocorrelations (I think tsa package can be useful for this aim). These summaries often give a guide to the ARMA model (p and q). On other hand, people say it is a good habit remove trend and seasonal components from its time series. Perhaps it is you need to remove 'peaks'. I do know detrend function remove trend but I don't know how remove seasonal component using prepackaged octave function (some methods are really easy, i.e.: http://www.nzmaths.co.nz/category/glossary/seasonal-component-time-series-data and everywhere in web...)

Cheers!

On Tue, Jan 3, 2012 at 17:34, miro99_ale <address@hidden> wrote:

Thank you Lukas... I appreciate your suggestion... :-)

Now, I have the last question: how I can identify a model (AR, MA, ARMA or
whatever) starting from the data? Is there a function in any package that
could approssimate my data to a model? How I could remove a peak in order to
not make worst the approximation?

--
View this message in context: http://octave.1599824.n4.nabble.com/Time-series-prediction-tp4247089p4258855.html

Sent from the Octave - General mailing list archive at Nabble.com.

_______________________________________________
Help-octave mailing list
address@hidden
https://mailman.cae.wisc.edu/listinfo/help-octave

-- 
«But Gwindor answered: 'The doom lies in yourself, not in your name.'»

JRR Tolkien