[FieldTrip] Error in Cluster permutation tutorial?

Robert Oostenveld r.oostenveld at donders.ru.nl
Mon Aug 17 15:20:59 CEST 2015


Hi Rajat and Stephen

Please see my inline reply below.


On 06 Aug 2015, at 16:32, Rajat Thomas wrote:

> I followed the tutorial on
> http://www.fieldtriptoolbox.org/tutorial/eventrelatedstatistics?s[]=cluster&s[]=permutation
> 
> at the very end, I wanted to do the ft_clusterplot but then I did not get topoplots as a function of time as in the
> tutorial and instead just got one plot.
> 
> Is there something wrong in the way the parameters are set in the tutorial. 

The code for ft_clusterplot was recently extended to be able to handle more types of data, but apparently it broke for the existing type of data. I have fixed it (see https://code.google.com/p/fieldtrip/source/detail?r=10603), it now supports chan, chan_time, chan_freq and chan_freq_time (but only if either time or freq has length 1). The fixed version is available from github or svn, or this evening on ftp. 


On 07 Aug 2015, at 16:43, Stephen Politzer-Ahles wrote:

> Also, is it just me or is http://www.fieldtriptoolbox.org/tutorial/eventrelatedstatistics#permutation_test_based_on_cluster_statistics essentially duplicating (with what looks like a more recent example and code) what's in http://www.fieldtriptoolbox.org/tutorial/cluster_permutation_timelock ? Would it be worthwhile to reconcile these two (and maybe replace one with a link to the other, so they don't get out of sync again)?

Content-wise they are similar, but we use the eventrelatedstatistics tutorial in training settings (e.g. workshops) as gentle introduction into statistics in general with only towards the end the clustering, whereas cluster_permutation_timelock goes more in depth. So they serve different goals. Since the code should now be working again for the 1st, I hope it also works for the 2nd.

best regards,
Robert







More information about the fieldtrip mailing list