leave-one-out cross validation and confusion matrix in knn

Question

I have to classify Iris data using k nearest neighbor, (k=1:30) I have divided the data into sample and training involving the Leave-one-out cross validation, so I have the following script:

load fisheriris
group=[ones(1,50), 2*ones(1,50), 3*ones(1,50)]';

    for k=(1:30);
         for i=(1:150);
         sample=meas(i,:);
         training1=meas;
         training1(i,:)=[];
         group_sample=group(i);
         group_training=group;
         group_training(i)=[];
         c(i,k)=knnclassify(sample,training1,group_training,k);

         A=confusionmat(group, c(i,k)); 
         mean_error(k)=mean(A(:)); 
         std_error(k)=std(A(:)); 
         end   
    end

The problem is that I can not make confusion matrix, since c returns me back only one value (for first sample), where is the problem, can anybody help?? Thanks!

Answer 1

I think you're probably after this:

for k=1:30

     for i=1:150
         sample=meas(i,:);
         training1=meas;
         training1(i,:)=[];
         group_sample=group(i);
         group_training=group;
         group_training(i)=[];
         c(i,k)=knnclassify(sample,training1,group_training,k);
     end   

     A=confusionmat(group, c(:,k)); 
     mean_error(k)=mean(A(:)); 
     std_error(k)=std(A(:)); 

end

So in other words only find the confusion matrix after the cross-validation loop.