How to divide dataset into a test, train, split format?

Hello,

I'm trying to split my dataset have the format X_train, X_test, y_train and y_test - in similar fashion to Python's test_train_split but I'm struggling to find a method to do so. Is this possible in MatLab?

I've tried doing the following

seed = 42;
rng(seed);
cv  = cvpartition(size(dataset,1), "HoldOut", 0.2);
idx = cv.test;
 
X_train = subsample(~idx,:);
y_test = subsample(idx,:);

but I'm not entirely sure how to go about deriving X_test and y_train.

Does anybody have a good solution to this? Apologies as I'm fairly new to MatLab!

Thank you!

Answer 1

[Ameer Hamza]

Does the variable subsample contains both 'X' and 'y' values? If yes, then you don't need to create two variables for X and 'y'. Just use

subsample_train = subsample(cv.training, :)
subsample_test = subsample(cv.test, :)

However, if subsample contains 'X' values and another variable (say, 'y') contain y values then you can do something like this

X_train = subsample(cv.training, :);
y_train = y(cv.training, :);
X_test = subsample(cv.test, :);
y_test = y(cv.test, :);

[Ziad]

Hey Ameer,

thanks for the reply!

I noticed my code had a typo, the correct code is:

seed = 42;
rng(seed);
cv  = cvpartition(size(subsample,1), "HoldOut", 0.2);
idx = cv.test;
 
X_train = subsample(~idx,:);
y_test = subsample(idx,:);

Could you eloborate more on your second point

However, if subsample contains 'X' values and another variable (say, 'y') contain y values then you can do something like this

as I'm not sure I understood it entirely.

Thank you!

[Ameer Hamza]

To explain the point, can you specify what data is stored in 'subsample'.

[Ziad]

Yes of course,

My subsample file contains my scaled, cleaned dataset containing my features and classes that I'll train my models on. It's a 664x31double.

[Ameer Hamza]

If first 30 columns are features and last column is label then you can do this

X_train = subsample(cv.training, 1:30);
y_train = y(cv.training, 31);
X_test = subsample(cv.test, 1:30);
y_test = y(cv.test, 31);

X_train and X_test are feature matrices and y_train and y_test are label vectors.

[Ziad]

Thank you very much, Ameer!

[Ameer Hamza]

I am glad to be of help!