[FieldTrip] 'avgoverchan' doesn't seem to do anything

[Eelke Spaak]

Hi Ryan,

Are you using ft_statistics_montecarlo directly? Or using one of the
high-level stats functions (e.g. ft_timelockstatistics, etc.)? It's
usually easier to use the latter, since they keep track of bookkeeping
stuff like averaging the data over various dimensions.

If you are using a high-level interface, have you tried specifying
cfg.neighbours = []?

Good luck,
Eelke

On Thu, 27 Feb 2020 at 21:22, Ryan Hubbard <rjhubba2 at illinois.edu> wrote:
>
> Hi all,
> I'm having an issue running cluster analyses on some data I have. Following this thread here (https://mailman.science.ru.nl/pipermail/fieldtrip/2013-July/006853.html), it seems like I should be able to specify 'avgoverchan' in my cfg, and then when I run a cluster analysis, it won't consider spatial information. In fact, looking at ft_statistics_montecarlo, I see a section that states:
>
>   elseif isfield(cfg, 'avgoverchan') && istrue(cfg.avgoverchan)
>       % channel dimension has been averaged across, no sense in clustering across space
>       cfg.connectivity = true(1);
>
> However, once we reach the "clusterstat" function this appears to be completely disregarded, and it searches for spatial information. I don't even see anything that reads the "connectivity" from the cfg in that function. I then get an error that states 'invalid dimension of spatdimneighbstructmat', as it is trying to look for channel neighbors to form clusters. This is true for both timelockstatistics and freqstatistics (as they both call ft_statistics_montecarlo).
>
> This is unfortunate, because it would be good to apply these cluster analysis functions to other time-series data (e.g. classification accuracies across time) where there aren't channels, but temporal clustering would still be useful. Can anyone help me with this?
>
> Thanks,
> Ryan
>
> --
> Ryan Hubbard
> Beckman Postdoctoral Fellow
> The Beckman Institute for Advanced Science and Technology
> University of Illinois at Urbana-Champaign
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202