[Eeglablist] Problems with the whiteness of the residuals

[Makoto Miyakoshi]

Dear John,

Honestly speaking, these things are quite mysterious to me, but looks like
AIC always suggest very high orders, which Tim said not very good and
should follow other suggestions... my problem is that I tend to end up with
using model order==30 all the time.

Makoto


2014-03-24 20:08 GMT-07:00 jfochoaster . <jfochoaster at gmail.com>:

> Thanks Makoto and Bethel
>
> I used another criteria for order selection, the AIC, and obtained higher
> orders and better results in the validation of the models
>
> Best wishes
>
>
> On Mon, Mar 24, 2014 at 7:17 PM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu>wrote:
>
>> Dear Bethel and John,
>>
>> I also often encounter this problem. It seems there is no guaranteed
>> solution for this. I almost always use order == 30 with piecewise
>> detrending on. Maybe you want to decrease the ICs to include (which could
>> be against your experimental interest, I know...)
>>
>> Should we contact Tim?
>>
>> Makoto
>>
>>
>>
>>
>> 2014-03-19 15:35 GMT-07:00 Bethel Osuagwu <b.osuagwu.1 at research.gla.ac.uk
>> >:
>>
>>> Hi
>>> What model order and window size are you using? You might need to give
>>> more details about your pre-processing steps including whether you are
>>> working with IC or sensors and the use of spatial filters. If your model
>>> order is too low for a given window size then your data will not be modeled
>>> properly and the residuals will not be white. So you might try using higher
>>> model orders but without over fitting.
>>>
>>> Bethel
>>> ________________________________________
>>> From: eeglablist-bounces at sccn.ucsd.edu [eeglablist-bounces at sccn.ucsd.edu]
>>> On Behalf Of jfochoaster . [jfochoaster at gmail.com]
>>> Sent: 18 March 2014 03:30
>>> To: eeglablist
>>> Subject: [Eeglablist] Problems with the whiteness of the residuals
>>>
>>> eeglab experts,
>>>
>>> I am working with ECoG data in eeglab using the SIFT plugin. When I am
>>> validating the model I obtain good consistency, above the 90%, good model
>>> stabilty, 100% stable, but bad whiteness of the residuals, none of the
>>> windows are white
>>>
>>> What I can do with these results?
>>>
>>> My data are adquired with the subject in resting, 240 Hz of sampling
>>> frequency, 16 channels, around 80 epochs of 2 seconds
>>>
>>> I really appreciate any help you can provide
>>>
>>> Best,
>>> --
>>> John Ochoa
>>> Docente de Bioingeniería
>>> Universidad de Antioquia
>>> _______________________________________________
>>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>>> To unsubscribe, send an empty email to
>>> eeglablist-unsubscribe at sccn.ucsd.edu
>>> For digest mode, send an email with the subject "set digest mime" to
>>> eeglablist-request at sccn.ucsd.edu
>>>
>>
>>
>>
>> --
>> Makoto Miyakoshi
>> Swartz Center for Computational Neuroscience
>> Institute for Neural Computation, University of California San Diego
>>
>
>
>
> --
> John Ochoa
> Docente de Bioingeniería
> Universidad de Antioquia
>



-- 
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20140327/7442c419/attachment.html>