How to filter first fixation times for participants

Edit: separated subjects and participants before attempted for loop, and removed unnecessary and obsolete information

Hello,

I'm trying to filtrate for the first fixations per participant per trial for eyelink data in Matlab. I have been able to separate pilot and subject trials as below

filename='RES_IAS (1).xlsx';                %Renamed file as filename
rawdata=readtable('filename')               %readcell undefined, used readtable instead 
idxPilot = startsWith(lower(rawdata.ID),'pilot'); %index for ID 'pilot'
idxSubj = startsWith(lower(rawdata.ID),'subj');   %index for ID 'subject' 
subjects_only = rawdata(idxSubj,:);     %extract data only for subjects
pilots_only = rawdata(idxPilot,:);      %extract data only for pilot

I then used sort rows to separate data that had fixation times, and then sorted again for the trial index to see when the first fixation times were.

subjects_only_sort= sortrows(subjects_only,{'IA_FIRST_FIXATION_TIME','ID'},{'ascend','descend'})
first_fixations = (subjects_only_sort(4006:end,:));  %extract only data with a fixation time > 0 
subject_sort = sortrows(first_fixations,{'TRIAL_INDEX'},{'ascend'}); %srearrange to trial ID

So to try and filter for the first fixations I created an empty array and a suggested for loop to run them through

fixations = {}
for k = 1:numel(subject_sort)
    if subject_sort.IA_FIRST_FIXATION_TIME > 0
        first = subject_sort.TRIAL_INDEX==1          
        if subject_sort.TRIAL_INDEX == first
            fixations = {fixations + first}
        end
    end
end
   

But I retrieve a line of several 0s, and my end array is still empty. What am I doing wrong?

Answer 1

[Peter Perkins]

Like SS, I tried to make sense of this code. And I found the same thing: in a recent version of MATLAB, these few lines would have done the job:

>> t = readtable("RES_IAS (1).xlsx");
Warning: Column headers from the file were modified to make them valid MATLAB identifiers before creating variable names for the table.
The original column headers are saved in the VariableDescriptions property.
Set 'VariableNamingRule' to 'preserve' to use the original column headers as table variable names. 
>> idx = (t.IA_FIRST_SACCADE_START_TIME > t.time);
>> scatter(t.IA_FIRST_SACCADE_START_TIME(idx),t.time(idx))
>> xlabel('time stimulus presented (ms)'), ylabel('time saccade seen(ms)'), title('Filtered Trials')

As for the loop at the end, SS took a shot, but the original code is so broken that it's hard to know what it is supposed to do. It's almost certain that it can be rewritten with no loop at all, but I don't understand what it's intended to do.

I also see at least three variables in that table that could usefully be converted to categorical, which may make you life down the road more convenient.

[Ruth Ronalter]

I'm using the MATLAB R2018a academic use. Unfortunately in this version, 'VariableNamingRule' is an undefined parameter, and scatter when used in the above context produces an 'invalid use of operator' error. Scatter also doesn't produce a figure unless called first. the best I can do is this

t = readtable("RES_IAS (1).xlsx");
idx = (t.IA_FIRST_SACCADE_START_TIME > t.time);
x = t.IA_FIRST_SACCADE_START_TIME(idx);     
y = t.time(idx);
figure;
scatter(x, y) %technically, I can rename the variables to anything else, but that's extra work
xlabel('time stimulus presented (ms)'), ylabel('time saccade seen(ms)'), title('Filtered Trials')

Apologies if this wasn't clear before, I guess Mathworks doesn't show that specific info when posting in the forums. As for what I'm trying to get the for loop to do, I was trying to get the first fixation per trial per subject, I thought the for loop would be able to do that, but it's proving not to be able to. Do you have a preferred method of filtering for a first entry with multiple trials? I have about 500 trials and 15 subjects

[Peter Perkins]

If you had timestamped data, you'd use retime with 'firstvalue' aggregation. Perhaps you can use rowfun with something like @(x) x(1).

[Ruth Ronalter]

Actually I do have data for the first fixation times, I'm trying the setup

retime(first_fix, 'regular', methods,'firstvalue', 'subjects')

but I get' Error using methods;Not enough input arguments.

If I don't use methods, I get "Undefined function 'retime' for input arguments of type 'table'"

Do I need to use timetable or table2timetable first?

[Peter Perkins]

Yes, retime is a method for timetables. Won't work on a table.