lcv

PURPOSE

SYNOPSIS

function g=lcv(varargin)

DESCRIPTION

 likelihood cross-validation.

 returned vector consists of log-likelihood, trace(H), trace(H'H)
 and cross-validated deviance (-2 * cross-validated log-likelihood).

 Author: Catherine Loader.

CROSS-REFERENCE INFORMATION

This function calls:

locfit Smoothing noisy data using Local Regression and Likelihood.
rsum

This function is called by:

lcvplot

SOURCE CODE

0001 function g=lcv(varargin)
0002 %
0003 % likelihood cross-validation.
0004 %
0005 % returned vector consists of log-likelihood, trace(H), trace(H'H)
0006 % and cross-validated deviance (-2 * cross-validated log-likelihood).
0007 %
0008 % Author: Catherine Loader.
0009 
0010 fit = locfit(varargin{:},'ev','cros');
0011 rs = rsum(fit);
0012 
0013 df0 = rs(1);
0014 df1 = rs(2);
0015 llk = rs(3);
0016 
0017 g = [llk df0 df1 -2*llk];
0018 
0019 return;