When you sort descending, the NaN values are actually sorted on top. You can use indexing to ignore NaN's before sorting. I'm adding a bit of extra code to keep track of the original position of these best 15 scores (you might want it later).
myData = rand(70,1);
myData([1,2,4,6]) = NaN;
idxKeepers = find(~isnan(myData));
[~,idxSort] = sort(myData(idxKeepers),'descend');
idxBestScores = idxKeepers(idxSort(1:15));
bestScores = myData(idxBestScores);
idxBestScores contains the location of the best 15 scores in your original data. bestScores contains the 15 best scores (sorted descending).
Best Answer