Local Regression and Likelihood, Figure 10.1. uses 'module','allcf' rather than 'deriv',[1 1] to get the second derivative coefficient Author: Catherine Loader
0001 % Local Regression and Likelihood, Figure 10.1. 0002 % 0003 % uses 'module','allcf' rather than 'deriv',[1 1] to get the 0004 % second derivative coefficient 0005 % 0006 % Author: Catherine Loader 0007 0008 load geyser; 0009 a = [0.5 0.75 1]; 0010 0011 f=['a','b','c','d']; 0012 for (i=1:3) 0013 figure('Name', sprintf( 'figure10_1%c', f(i) ) ); 0014 0015 fit = locfit(geyser,'deg',2,'link','ident','kern','gauss','ev','grid','mg',501,'ll',1,'ur',6,'module','allcf','h',a(i)); 0016 x = fit.fit_points.evaluation_points; 0017 y = lfknots(fit); 0018 y = y(:,3); 0019 plot(x,y.^2); 0020 itgl = [0.5 ones(1,499) 0.5] * (y.^2) / 100 0021 end; 0022