m2.m

粒子群算法

N=100;noise=0.5;
X= 10*[-1:2/(N-1):1]';
t= sin(X).*exp(-0.2*X);
Y=t+ noise*randn(N,1);
model = svmtrain(Y, X, '-s 3 -t 2 -g 0.32 -c 10 -p 0.1 -b 1 -p 1');
[y, accuracy] = svmpredict(Y, X, model);
figure(3)
clf;
h_y=plot(X,t,'r--','LineWidth',2);
hold on
plot(X,Y,'g.');
hold on
 h_ysvm=plot(X,y,'b','linewidth',2);
  %legend([h_y h_ysvm],'initial function','SVM predictor')
    fprintf('\nSVM regression test error (RMSE): %g\n', ...
	sqrt(mean((t-Y).^2)))   
    fprintf('\nSVM regression test error (MAXE): %g\n', ...
    max(abs(t-Y)))