That's the kind of thing I was expecting to happen. The problem is
S.WeightAve = 1 setting. If you weight by the number of replication
and one of those numbers is zero you will get NaNs. So you should set
S.WeightAve = 0. That would not be suitable for combining several
conditions into one so perhaps you should have two steps in one of
which you combine all the conditions you want to combine with
weighting and in the next step you compute the differences without
weighting.
Vladimir
On Tue, Mar 20, 2012 at 4:19 PM, Muhammad Adeel Parvaz
<[log in to unmask]> wrote:
> Yes. I think I know what's going on here. After computing averages for each
> condition (and inserting conditions with nrepl of 0), I am also creating
> some higher level contrasts, such as averaging different conditions,
> subtracting a condition from the other and etc. This is where the NaNs are
> creeping in for the condition which do not have any events in there. The
> excerpt from the script is as follows:
>
> %%
> % Adding missing conditions into right order (suggested by V.Litvak)
> corig = D.condlist;
>
> c = zeros(numel(cnew), numel(corig));
> [sel1, sel2] = spm_match_str(cnew, corig);
>
> for i = 1:length(sel1)
> c(sel1(i), sel2(i)) = 1;
> end
>
> nrepl = zeros(1, numel(cnew));
> nrepl(sel1) = D.repl;
> %%
> S = [];
> S.D = D;
> S.c = c;
> S.label = cnew;
> S.WeightAve = 0;
> D = spm_eeg_weight_epochs(S);
> %%
> delete(S.D);
> D = repl(D, [], nrepl);
>
> %%
> % Create Relevant contrasts (THESE ARE THE HIGHER LEVEL CONTRASTS I AM
> TALKING ABOUT)
> S = [];
> S.D = D;
> S.c = [
> 1 0 0 0 0 0 0 0 0 0 0 0
> 0 1 0 0 0 0 0 0 0 0 0 0
> 0 0 1 0 0 0 0 0 0 0 0 0
> 0 0 0 1 0 0 0 0 0 0 0 0
> 0 0 0 0 1 0 0 0 0 0 0 0
> 0 0 0 0 0 1 0 0 0 0 0 0
> 0 0 0 0 0 0 1 0 0 0 0 0
> 0 0 0 0 0 0 0 1 0 0 0 0
> 0 0 0 0 0 0 0 0 1 0 0 0
> 0 0 0 0 0 0 0 0 0 1 0 0
> 0 0 0 0 0 0 0 0 0 0 1 0
> 0 0 0 0 0 0 0 0 0 0 0 1
> 1 0 0 0 1 0 0 0 1 0 0 0
> 0 1 0 0 0 1 0 0 0 1 0 0
> 0 0 1 0 0 0 1 0 0 0 1 0
> 0 0 0 1 0 0 0 1 0 0 0 1
> 1 1 0 0 1 1 0 0 1 1 0 0
> 0 0 1 1 0 0 1 1 0 0 1 1
> -1 0 1 0 -1 0 1 0 -1 0 1 0
> 0 -1 0 1 0 -1 0 1 0 -1 0 1
> -1 -1 1 1 -1 -1 1 1 -1 -1 1 1];
> S.label = {
> '1-Cue Pred Win'
> '1-Cue Unpred Win'
> '1-Cue Pred Loss'
> '1-Cue Unpred Loss'
> '2-Cue Pred Win'
> '2-Cue Unpred Win'
> '2-Cue Pred Loss'
> '2-Cue Unpred Loss'
> '3-Cue Pred Win'
> '3-Cue Unpred Win'
> '3-Cue Pred Loss'
> '3-Cue Unpred Loss'
> 'Pred Win'
> 'UnPred Win'
> 'Pred Loss'
> 'UnPred Loss'
> 'All Win'
> 'All Loss'
> 'Pred LossMinusWin'
> 'Unpred LossMinusWin'
> 'All LossMinusWin'}';
> S.WeightAve = 1;
> D = spm_eeg_weight_epochs(S);
> %%
>
>
> This is where the problem is coming from. How do you suggest i do this and
> at the same time keep NaN away from creeping in?
>
> Thanks
> Muhammad
>
>
> On Tue, Mar 20, 2012 at 11:54 AM, Vladimir Litvak
> <[log in to unmask]> wrote:
>>
>> You should be able to view datasets with zeros. It's NaNs that are
>> problematic. So you should see at what stage in the processing you are
>> getting them. In principle I think if you switch to a condition that
>> does not contain NaNs it should plot OK.
>>
>> Vladimir
>>
>> On Tue, Mar 20, 2012 at 3:49 PM, Muhammad Adeel Parvaz
>> <[log in to unmask]> wrote:
>> > Just tried that. I could look at the individual averages of 4 out of 9
>> > subjects. Out of the 5, that I could not view the averages, 4 had one or
>> > more conditions with 0 events, and 1 had data for all conditions. Why do
>> > you
>> > think I am not able to view the data for this subject. Also, how can I
>> > view
>> > the data (atleast for the conditions that had non-zero events) for the
>> > other
>> > 4 subjects?
>> >
>> > Thanks
>> > Muhammad
>> >
>> > On Tue, Mar 20, 2012 at 11:36 AM, Vladimir Litvak
>> > <[log in to unmask]> wrote:
>> >>
>> >> You should make sure you can plot all the individual averages. Then
>> >> you can try averaging just two subjects then 3 etc. and find where it
>> >> breaks. Maybe just one subject is problematic for some reason.
>> >>
>> >> Best,
>> >>
>> >> Vladimir
>> >>
>> >> On Tue, Mar 20, 2012 at 3:32 PM, Muhammad Adeel Parvaz
>> >> <[log in to unmask]> wrote:
>> >> > Thanks Vladimir for your prompt response. That was my first guess as
>> >> > well.
>> >> > But this is also happening, when I select only those subjects who had
>> >> > data
>> >> > in all conditions. So, I am not sure where these NaN are coming from.
>> >> > How di
>> >> > i check for this? Ideally, i would like to include all the subjects
>> >> > (including those with some missing conditions) in the grand average.
>> >> >
>> >> > Thanks
>> >> > Muhammad
>> >> >
>> >> > On Tue, Mar 20, 2012 at 11:22 AM, Vladimir Litvak
>> >> > <[log in to unmask]> wrote:
>> >> >>
>> >> >> Dear Muhammad,
>> >> >>
>> >> >> This might indicate that you have some NaNs in your data which in
>> >> >> turn
>> >> >> might be related to your missing condition problem and the fact that
>> >> >> in some cases you might have had division by zero.
>> >> >>
>> >> >> Best,
>> >> >>
>> >> >> Vladimir
>> >> >>
>> >> >> On Tue, Mar 20, 2012 at 3:06 PM, Muhammad Adeel Parvaz
>> >> >> <[log in to unmask]> wrote:
>> >> >> > Hello all,
>> >> >> >
>> >> >> > I have analyzed a few EEG subjects, and have created a grand
>> >> >> > average
>> >> >> > of
>> >> >> > the
>> >> >> > ERPs. However, after doing the grand averages, I get the following
>> >> >> > error
>> >> >> > when I try to look at the plots:
>> >> >> >
>> >> >> > ??? Error using ==> set
>> >> >> > Bad value for axes property: 'YLim'
>> >> >> > Values must be increasing and non-NaN.
>> >> >> >
>> >> >> > Error in ==> spm_eeg_review_callbacks>updateDisp at 1182
>> >> >> > set(handles.axes(i),'color',color,...
>> >> >> >
>> >> >> > Error in ==> spm_eeg_review_callbacks at 723
>> >> >> > updateDisp(D,1)
>> >> >> >
>> >> >> > ??? Error while evaluating uicontrol Callback
>> >> >> >
>> >> >> > Can anyone please explain why plotting function is unable to set
>> >> >> > 'YLim'
>> >> >> > and
>> >> >> > how can I check if the values are indeed increasing and are not
>> >> >> > non-NaN,
>> >> >> > as
>> >> >> > suggested by the error?
>> >> >> >
>> >> >> > Many thanks,
>> >> >> > Muhammad
>> >> >> >
>> >> >
>> >> >
>> >
>> >
>
>
|
