Dear Chih-Hao Lien,
I don't know if Joram is monitoring this list so best would be to open
an issue in his GitHub repository:
https://github.com/JoramSoch/MACS/issues
Best regards,
Guillaume.
On 23/04/2021 11:58, Lien Chih-hao wrote:
> Dear all,
>
> I'm new to use MACS toolbox (https://github.com/JoramSoch/MACS) to
> select the better model for my data.
> <https://github.com/JoramSoch/MACS>
>
> JoramSoch/MACS: MACS – a new SPM toolbox for model assessment,
> comparison and selection - GitHub <https://github.com/JoramSoch/MACS>
> MACS. MACS – a new SPM toolbox for model assessment, comparison and
> selection. This toolbox (pronounced as "Max") evaluates general linear
> models (GLMs) for functional magnetic resonance imaging (fMRI) data
> estimated in Statistical Parametric Mapping (SPM).
> github.com
>
>
> I try to compare models that have different confoundings (i.e. different
> head motion parameters and global signals)
> But I get some warning messages when doing Bayesian model selection.
>
> The warning is:
>
> Warning: Matrix is close to singular or badly scaled. Results may be
> inaccurate. RCOND = 1.119708e-16.
>
>
> After a few tests, I find that the warning happens when I include some
> subjects with the specific models (i.e. 36 head motion parameters).
>
> Does anyone meet the same situation?
>
> Best regards,
> Chih-Hao Lien
--
Guillaume Flandin, PhD
Wellcome Centre for Human Neuroimaging
UCL Queen Square Institute of Neurology
London WC1N 3BG
|