You may find the discussion in this thread useful.
http://book.caltech.edu/bookforum/showthread.php?t=413
Quote:
Originally Posted by melipone
For  , I get the variance of the hypotheses. But how do I compute the expected value of the out-of-sample error? Do I generate different data points and test each hypotheses against each of those points and take the mean square error? But I don't get 0.75 so I must be doing something wrong.
How do you compute the expected out-of-sample error for in Example 2.8? |