Hi,
In the Help example, the crossvalind.m is used once outside the for loop. What exactly does Indices contain, the result for one partitioning or the average of K iterations?
Do I need to call crossvalind inside a for loop in order to  esure that "This process is repeated K times, leaving one different fold for evaluation each time"? 

> In the Help example, the crossvalind.m is used once outside the for loop. 
> What exactly does Indices contain, the result for one partitioning or the 
> average of K iterations?

Just take a look:

   load fisheriris
   indices = crossvalind('Kfold',species,10);

You'll see that it's a numeric vector containing the fold number assigned to 
each row.

> Do I need to call crossvalind inside a for loop in order to  esure that 
> "This process is repeated K times, leaving one different fold for 
> evaluation each time"?

By leaving it outside the loop you insure that the assignment of rows to 
folds doesn't change. So then when you leave out one fold at a time in the 
loop, you're leaving out each observation exactly once.

-- Tom