|
wolffd@0
|
1 function [likXandY, likYgivenX, post] = cwr_prob(cwr, X, Y);
|
|
wolffd@0
|
2 % CWR_EVAL_PDF cluster weighted regression: evaluate likelihood of Y given X
|
|
wolffd@0
|
3 % function [likXandY, likYgivenX, post] = cwr_prob(cwr, X, Y);
|
|
wolffd@0
|
4 %
|
|
wolffd@0
|
5 % likXandY(t) = p(x(:,t), y(:,t))
|
|
wolffd@0
|
6 % likXgivenY(t) = p(x(:,t)| y(:,t))
|
|
wolffd@0
|
7 % post(c,t) = p(c | x(:,t), y(:,t))
|
|
wolffd@0
|
8
|
|
wolffd@0
|
9 [nx N] = size(X);
|
|
wolffd@0
|
10 nc = length(cwr.priorC);
|
|
wolffd@0
|
11
|
|
wolffd@0
|
12 if nc == 1
|
|
wolffd@0
|
13 [mu, Sigma] = cwr_predict(cwr, X);
|
|
wolffd@0
|
14 likY = gaussian_prob(Y, mu, Sigma);
|
|
wolffd@0
|
15 likXandY = likY;
|
|
wolffd@0
|
16 likYgivenX = likY;
|
|
wolffd@0
|
17 post = ones(1,N);
|
|
wolffd@0
|
18 return;
|
|
wolffd@0
|
19 end
|
|
wolffd@0
|
20
|
|
wolffd@0
|
21
|
|
wolffd@0
|
22 % likY(c,t) = p(y(:,t) | c)
|
|
wolffd@0
|
23 likY = clg_prob(X, Y, cwr.muY, cwr.SigmaY, cwr.weightsY);
|
|
wolffd@0
|
24
|
|
wolffd@0
|
25 % likX(c,t) = p(x(:,t) | c)
|
|
wolffd@0
|
26 [junk, likX] = mixgauss_prob(X, cwr.muX, cwr.SigmaX);
|
|
wolffd@0
|
27 likX = squeeze(likX);
|
|
wolffd@0
|
28
|
|
wolffd@0
|
29 % prior(c,t) = p(c)
|
|
wolffd@0
|
30 prior = repmat(cwr.priorC(:), 1, N);
|
|
wolffd@0
|
31
|
|
wolffd@0
|
32 post = likX .* likY .* prior;
|
|
wolffd@0
|
33 likXandY = sum(post, 1);
|
|
wolffd@0
|
34 post = post ./ repmat(likXandY, nc, 1);
|
|
wolffd@0
|
35 %loglik = sum(log(lik));
|
|
wolffd@0
|
36 %loglik = log(lik);
|
|
wolffd@0
|
37
|
|
wolffd@0
|
38 likX = sum(likX .* prior, 1);
|
|
wolffd@0
|
39 likYgivenX = likXandY ./ likX;
|
