Expression or statement is incorrect--possibly unbalanced (, {, or [.

hello every body i have an error ??? Error: File: test.m Line: 31 Column: 3 exactely in [~,scores] = predict(cl,xGrid); i have Matlab 7.8.0 (R2009a)

1. rand(1); % For reproducibility
2. r = sqrt(rand(100,1)); % Radius
3. t = 2*pi*rand(100,1); % Angle
4. data1 = [r.*cos(t), r.*sin(t)]; % Points
6. r2 = sqrt(3*rand(100,1)+1); % Radius
7. t2 = 2*pi*rand(100,1); % Angle
8. data2 = [r2.*cos(t2), r2.*sin(t2)]; % points
10. figure;
11. plot(data1(:,1),data1(:,2),'r.','MarkerSize',15)
12. hold on
13. plot(data2(:,1),data2(:,2),'b.','MarkerSize',15)
14. ezpolar(@(x)1);ezpolar(@(x)2);
15. axis equal
16. hold off
18. data3 = [data1;data2];
19. theclass = ones(200,1);
20. theclass(1:100) = -1;
22. %Train the SVM Classifier
23. cl = fitcsvm(data3,theclass,'KernelFunction','rbf',...
24. 'BoxConstraint',Inf,'ClassNames',[-1,1]);
26. % Predict scores over the grid
27. d = 0.02;
28. [x1Grid,x2Grid] = meshgrid(min(data3(:,1)):d:max(data3(:,1)),...
29. min(data3(:,2)):d:max(data3(:,2)));
30. xGrid = [x1Grid(:),x2Grid(:)];
31. [~,scores] = predict(cl,xGrid);
33. % Plot the data and the decision boundary
34. figure;
35. h(1:2) = gscatter(data3(:,1),data3(:,2),theclass,'rb','.');
36. hold on
37. ezpolar(@(x)1);
38. h(3) = plot(data3(cl.IsSupportVector,1),data3(cl.IsSupportVector,2),'ko');
39. contour(x1Grid,x2Grid,reshape(scores(:,2),size(x1Grid)),[0 0],'k');
40. legend(h,{'-1','+1','Support Vectors'});
41. axis equal
42. hold off

[KSSV]

It is working pretty fine in 2017.

Answer 1

[Steven Lord]

The ability to ignore specific input or output arguments in function calls using the tilde operator was introduced in release R2009b. Replace ~ with a dummy variable name, like dummy, for older releases.

[ghali ahmed]

i'll to thank you for quick response

i have an other problem

??? Undefined function or method 'fitcsvm' for input arguments of type 'double'.
Error in ==> test at 23
cl = fitcsvm(data3,theclass,'KernelFunction','rbf',...

[per isakson]

fitcsvm - Train binary support vector machine classifier
fitcsvm trains or cross-validates a support vector machine (SVM)
model for two-class (binary) classification on a low- through 
moderate-dimensional predictor data set. fitcsvm supports...

Documentation > Statistics and Machine Learning Toolbox > Classification > Support Vector Machine Classification

[Walter Roberson]

That routine was introduced in R2014a.

In your software release there was no built-in SVM in any toolbox, so people would compile and link the third party libsvm