fitcsvm error - "only floating-point data for x"

Hi, I am trying to train my SVM model (emotion classification using EEG).

I get several errors from the below line of code:

>>SVMmodelA=fitcsvm(D,Abin,'KernelFunction','Linear','Standardize',true,'ClassNames',{'low arousal','high arousal'})

Error using ClassificationSVM.prepareData (line 618)
You can pass only floating-point data for X to SVM.

Error in classreg.learning.FitTemplate/fit (line 213)
                    this.PrepareData(X,Y,this.BaseFitObjectArgs{:});

Error in ClassificationSVM.fit (line 239)
            this = fit(temp,X,Y);

Error in fitcsvm (line 278)
obj = ClassificationSVM.fit(X,Y,varargin{:});

________________

Re the first error, another post on here says convert your array into a matrix which I have already done.

Can anyone help??

Are there any other functions or ways of training an SVM?

[Julia Down]

This was the answer: SVMmodelA=fitcsvm(T.data,Abin,'KernelFunction','Linear','Standardize',true,'ClassNames',{'0','1'})

[Walter Roberson]

Yup, that would do it.

Answer 1

[Walter Roberson]

SVMmodelA = fitcsvm(double(D), double(Abin), 'KernelFunction', 'Linear', 'Standardize', true, 'ClassNames', {'low arousal', 'high arousal'})

[Julia Down]

Hi Walter, thank you for your answer. What does using double on the data and label attribute do?

I did it and got this error: Error using double Conversion to double from struct is not possible.

Thanks

[Walter Roberson]

Which MATLAB version are you using?

Please show how D and Abin are constructed and what their class() and size() are.

fitcsvm() accepts only table objects or a numeric array for its first argument, but it appears you are passing in a struct for the first argument.

Answer 2

[ALi Arshad]

Hi, I am trying to train my SVM model i am getting below error

Error in fitcsvm (line 278) obj = ClassificationSVM.fit(X,Y,varargin{:});

Error in ali3 (line 8) SVMMod = fitcsvm(Data1, ClassName);

[Walter Roberson]

What is class(Data1) and class(ClassName) and the size() of each of those?