You must pass X as a floating-point matrix.

I am building a GUI to classify breast images. I am using SVM as the classifier technique. However, I am getting an error message. Please find my code an the error message below. Any help would be appreciated.

File=handles.File;
InputImage=handles.InputImage;
TestSet=InputImage;
Labels = table2array(File);
Training=Labels(1:2004,1:9);
class=Labels(:,10);
SVMmodel= fitcsvm(Training, class, 'KernelFunction', 'Linear', 'Standardize', true, 'ClassNames', {'1', '2'});
result = predict(SVMmodel, TestSet);
result=num2str(result);

The error message is displayed below:

Error using classreg.learning.impl.CompactSVMImpl/score (line 45)

You must pass X as a floating-point matrix.

Error in classreg.learning.classif.CompactClassificationSVM/score (line 591)

f = score(this.Impl,X,true,varargin{:});

Error in classreg.learning.classif.ClassificationModel/predict (line 411)

scores = score(this,X,varargin{:});

Error in classreg.learning.classif.CompactClassificationSVM/predict (line 433)

[e-mail](this,X,varargin{:});

Error in new>pushbutton4_Callback (line 143)

result = predict(SVMmodel, TestSet);

Error in gui_mainfcn (line 95)

feval(varargin{:});

Error in new (line 42)

gui_mainfcn(gui_State, varargin{:});

Error in matlab.graphics.internal.figfile.FigFile/read>@(hObject,eventdata)new('pushbutton4_Callback',hObject,eventdata,guidata(hObject))

Error while evaluating UIControl Callback.

[Athrey Ranjith Krishnanunni]

From the documentation for predict, it says that the syntax is

predict(Mdl,X)

where X is the predictor data, and should be a numeric array.

In your case, X is TestSet, so try running

whos('TestSet')

in the command line to see what comes up under the Size and Class headings.

Answer 1

[Ive J]

Your TestSet must have the same structure as your Training set. You can try this

result = predict(SVMmodel, Labels(:, 1:9));

[Warid Islam]

Hi Ive,

Thank you for the answer. It worked. I have another query. I am applying SVM to classify images. If the result variable is 1, then it should display Malignant in the edit box. Otherwise it should display Benign. I tried the code below, but it didn't work.

SVMmodel= fitcsvm(Training, class, 'KernelFunction', 'Linear', 'Standardize', true, 'ClassNames', {'1', '2'});
result = predict(SVMmodel, TestSet);
if result==1
    ac='Malignant';
else
    ac='Benign';
end
set(handles.edit1,'String',ac);

The following error message is displayed.

Undefined operator '==' for input arguments of type 'cell'.

Error in new>pushbutton4_Callback (line 163)

if result==1

Error in gui_mainfcn (line 95)

feval(varargin{:});

Error in new (line 42)

gui_mainfcn(gui_State, varargin{:});

Error in matlab.graphics.internal.figfile.FigFile/read>@(hObject,eventdata)new('pushbutton4_Callback',hObject,eventdata,guidata(hObject))

Error while evaluating UIControl Callback.

[Walter Roberson]

The return value from predict is labels in the same format as they were input to ficsvm. Your two labels are {'1', '2'} so you get a cell array of labels returned.

Consider changing the {'1', '2'} to be {'Malignant', 'Benign'} and then you would not have to do the if .

[Warid Islam]

Hi Walter,

Your suggestions worked big time. Thank you.

Best Regards

Warid Islam