Back to the main page.

Bug 651 - the prepare_design function should be deprecated/removed

Reported 2011-05-10 20:58:00 +0200
Modified 2012-08-03 14:53:44 +0200
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Mac OS
Importance: P1 normal
Assigned to: Roemer van der Meij
Depends on:
See also:

Robert Oostenveld - 2011-05-10 20:58:35 +0200

it is located in private and used to create a design on the fly. Instead of automatically creating (a potentially incorrect) design, the user should be forced to think about it. All (wiki) documentation should be checked. All consequences should be examined. An email should be sent to the list.

Robert Oostenveld - 2011-05-11 14:24:26 +0200

also look for a potential neighbourselection bug and fix that one as well

Johanna - 2011-05-11 15:37:36 +0200

neighbourselection is in bug 591.

Johanna - 2011-05-11 15:42:01 +0200

note to self: check all other bugs pertaining to ft_neighbourselection

Johanna - 2011-05-16 11:38:36 +0200

In ft_sourcestatistics (line 286) and statistics_wrapper (line 248): may also exist. Should user put design in cfg or data, or either ok? (If put in both, make sure it is same design, or let one override?) More needed on wiki or in help-comments of two fcns above, than just for how to create a design matrix (i.e. what does this matrix mean, depending on what stats you are doing). Create FAQ for this? 'how to create a design matrix?' Is it always a 1xN vector assigning a condition to a trial, or can it be MxN where M are different ways of describing trials (e.g. left/right response, correct/incorrect response, cue probability, etc )? If so, how does one choose which are factors-of-interest and factors-of-no-interest?

Johanna - 2011-09-14 14:42:01 +0200

thanks to Robert for creating:

Roemer van der Meij - 2012-03-21 11:17:05 +0100

Note to self: currently called by:: grep 'prepare_design' *.m ./*/*.m ./*/*/*.m ./*/*/*/*.m -l ft_sourcestatistics.m (called on absence of ./private/statistics_wrapper.m (called on absence of Replace these by an error stating that the should be specified, with a short reference to the FAQ on the wiki.

Roemer van der Meij - 2012-08-03 14:53:44 +0200

The information on the design matrix is a bit scattered throughout the documentation. It is for example mentioned in the doc of freqstatistics, but not in timelockstatistics or sourcestatistics. There is a good description in the tutorials for event-related and tfr data, but not for source data? There is a good faq on it, but from the name it looks like it only pertains to random permutations, while ft_statistics_analytic also needs it. How specific is it for all the stat-functions and data-types? Is the definition generic enough to simply put it in one location? Or do we make a FAQ for every stat/data-type?