crossval leaveout for SVM

Hi, I'm trying to code a leave one subject out cross validation after training my data with SVM. There are 15 subjects. I've read the crossval documentation but am still lost... Here is my code:

load EEG_X  %features
load EEG_Y  %label
Subject_index=2;
fea = X{1,Subject_index};
gnd = Y{1,Subject_index};
model = train(gnd', sparse(fea'));
c = cvpartition(15,'leaveout')
cvmodel = crossval('CVPartition',c)

I'm not sure about the relationship between cvpartition and crossval. Any help would be greatly appreciated!