Splitting data using loop

Hello, I have data of 60users(each user data has different no. Of rows but same no. Of columns)and I have to split each user data into training and test set in 60,40% randomly and then combine all training data(of 60 users)in one matrix and all test data(of 60 users)in another matrix. I am using dividerand to split for each user. Can anyone please suggest how to use loop to do this task efficiently?

Thank you.

[SBS]

Any help please?

Answer 1

[Jalaj Gambhir]

Hi,

This can be achieved using findgroups and splitapply. For the example given below, I have used fisherirs dataset, which contains 3 categories of flowers (can be extended to 60 users) and 50 samples for each category in the dataset. These 50 samples have been split to 60% train and 40% test.

Hope this helps!

load fisheriris;
groups = findgroups(species);
func = @(x) {x};
grouped_data = splitapply(func, meas, groups);
train_data = [];
test_data = [];
for category = 1:length(grouped_data)
    train_percent = 0.6
   [rows,col] = size(grouped_data{category});
   idx = randperm(rows);
   training_i = grouped_data{category}(idx(1:round(train_percent*rows)),:);
   testing_i = grouped_data{category}(idx(round(train_percent*rows)+1:end),:);
   train_data = vertcat(train_data,training_i);
   test_data = vertcat(test_data,testing_i);
end