Mercurial > repos > vipints > rdiff
comparison rDiff/src/locfit/m/kappa0.m @ 0:0f80a5141704
version 0.3 uploaded
author | vipints |
---|---|
date | Thu, 14 Feb 2013 23:38:36 -0500 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:0f80a5141704 |
---|---|
1 function kap=kappa0(x,y,varargin) | |
2 | |
3 % Compute the constants for `tube-formula' based simultaneous | |
4 % confidence bands. | |
5 % | |
6 % Works for regression models only. Density estimation problems | |
7 % should be converted to counts, and use poisson regression | |
8 % 'family','poisson'. | |
9 % | |
10 % Essentially, this is a front-end to locfit, and so all optional | |
11 % arguments to locfit (eg, smoothing parameters) can be provided. | |
12 % | |
13 % To compute (or plot) the confidence bands, provide the output | |
14 % of the kappa0() function as the 'kappa' argument to a | |
15 % predict() or lfband() call. | |
16 % | |
17 % | |
18 % Example: | |
19 % | |
20 % load ethanol; | |
21 % fit = locfit(E,NOx,'alpha',0.5) | |
22 % kap = kappa0(E,NOx,'alpha',0.5) % give same arguments! | |
23 % lfplot(fit) | |
24 % lfband(fit,'kappa',kap) % plot the simultaneous bands | |
25 % z = predict(fit,[0.6 0.7 0.8]','kappa',kap,'band','g') | |
26 % z{3} % evaluate the bands. | |
27 | |
28 fit = locfit(x,y,'module','kappa','ev','grid','mg',20,varargin{:}); | |
29 z = fit.fit_points.kappa; | |
30 d = size(fit.data.x,2); | |
31 kap = z(1:(d+1)); | |
32 | |
33 return; |