How to do multiple label classification using libSVM

Itban Saeed

2016 12 6

0 个回答

1 次查看（30 天）

0 个投票

I am a beginner of MATLAB and to be honest, quite weak in mathematics. I am trying to use libSVM to classify data between multiple classes. I am reading the data from `xlsx` file and following this code but *the issue is that svmpredict function takes two double vectors and I need to process one text vector (labels) and one double matrix (data)*. Basically I want to convert the following SVM code to libSVM (for multiple classes)

*SVM Code:*

    %%Read the .xlsx file and build SVM model for English and Urdu data
    name = strcat(fileName,'.xlsx');
    [English] = xlsread(name,1,'C:H');
    [Urdu] = xlsread(name,2,'C:H');
    [Test_data] = xlsread(name,3,'C:H');
    [~, labels1]=xlsread(name,3,'I:I');
    test_labels=labels1(2:end);
    labels=[1*ones(size(English,1),1) ; 2*ones(size(Urdu,1),1)];
    data = [English ; Urdu];
    SVMStruct = svmtrain(data,labels);
    %%Generate the output file
    % result will be an array containing 0 for the points
    % detected as Urdu ones and 1 for the English detected
    % points.
    result = svmclassify(SVMStruct,Test_data);

The above code works fine for two classes and now I want to classify between multiple classes. Someone please guide me about any function/tutorial etc to classify data between multiple classes.