For loop for SVM variable files

I want to classify my data with SVM on MATLAB. I have a folder, this folder has 70 sub-folders. Each of the 70 sub-folders has three preprocessed .mat files(selected features). These three files are my predictor variables. My design matrix is in a .mat file. Inside the design matrix, there are 3 items, out of which I want to use 2 as my class variables.

Using fitcsvm or any other Matlab svm function, I'm trying to classify that the predictor variables Fall into one of the two classes.

How do write a for loop that runs through these subfolders so that X = the predictor variables and Y= the class variables.

Answer 1

[Balaji Udayagiri]

Hi Monalisa

As per my understanding, you want to loop through the folders to read the predictor variables for the SVM.

D = dir; %folder that contains the 70 data folders 
D = D(~ismember({D.name}, {'.', '..'})); 
for i = 1:numel(D) 
    subD = D(i).name; %each data folder containing the 3 subfolders 
    for j = 1:numel(subD) 
        currD = subD(j).name;  %folder of the individual predictor variable 
        fList = dir(currD); % list contains the files in the current directory 
        
        % here you can read the data into your input variables, Not sure how the data is present in the folder.  
        
    end 
end 

[Monalisa Chikezie]

Hi @Balaji Udayagiri, thank you for the swift response. I updated the question. Does this still hold with the updated question?

Answer 2

[Pamudu Ranasinghe]

I think you can use "datastore" function in matlab

https://in.mathworks.com/help/matlab/ref/datastore.html

You dont need to loop through the files use 'IncludeSubfolders' to 'true' in datastore function.

This will do your work