Cross validation in Support vector machine (SVM).

2 次查看（过去 30 天）

Max 2015-1-27

2
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/172081-cross-validation-in-support-vector-machine-svm

Hello all,

I am trying to find out of sample mis-classification for my SVM model using by using:

CVSVMModel1 = crossval(SVMModel1);
misclass1 = kfoldLoss(CVSVMModel1);

This is from the link

http://www.mathworks.com/help/stats/support-vector-machines-svm.html#bsr5o09

However, I keep getting error as below:

Error using crossval (line 160)
At least two inputs are needed.

Since I just simply copied the codes from the link above, I assume the problem is perhaps due to MATLAB version! There was other errors that I had to use "svmtrain" instead of "fitcsvm". My MATLAB version is 2013b. Has anybody encountered similar problem? Could anybody give me a hint how to cross validate and finally get the out-of-sample mis-classification rate (please refer to the link above) on the trained ?