Prev: Image Processing Help required...
Next: FIFOfor HDL code generation
From: Xander on 7 Jun 2010 02:03
I am trying to fit a single ARMAX/GARCH model to multiple time series to produce a single model that is the best fit to the aggregate signal of all the time series. Obviously I cannot simply append one time series to the next since the transition between series would produce errors in the lag operators. Likewise, I can't pad the transitions zones for the length of the longest lag since it would introduce anomolies as well. Dissection of garchfit() makes the job of customizing how the routine would need to calculate this aggregate model doesn't seem like a straight forward problem, and I was hoping that someone might have attempted this and would have any suggestions.

Thanks,
Xander

PS My understanding of VARMAX is such that it doesn't look as though it would fit the bill here, but if someone sees it otherwise please let me know
From: Oscar Olarte on 15 Jun 2010 11:17
Why not make a model for each signal and then an average model?
 | 
Pages: 1
Prev: Image Processing Help required...
Next: FIFOfor HDL code generation