How to randomly select data out of a dataset?

50 views (last 30 days)
Ines on 29 May 2012
hey there, I want to randomly select 80% from my data to create a training dataset and use the residual 20% for the evaluation of my model obtained from the training dataset. How I can I best perform this split in matlab? (Actually I want to perform this split multiple times within a loop in order to be able to deliver a more robust result)

Answers (3)

Wayne King
Wayne King on 29 May 2012
One way if you have the Statistics Toolbox is to use randsample
x = randn(1000,1);
y = randsample(length(x),800);
Another way if you don't have the Statistics Toolbox.
R = randperm(length(x));
indices = R(1:800);
y = x(indices);
  1 Comment
Peter Perkins
Peter Perkins on 29 May 2012
In newer release of the Statistics Toolbox, you can/should use datasample, rather than randsample. It does some things better, and is perhaps a little easier to use.

Sign in to comment.

Thomas on 29 May 2012
I dont know if this will help..
Suppose your data is in a
a=rand(10,1); % generate random data
[trainingset]= intersect(a,randsample(a,8)) % gives training set with 8 random samples from a you can set what size your trainign set needs to be
testset=a(~ismember(a,trainingset)) % gives test set

Peter Perkins
Peter Perkins on 29 May 2012
Another possibility if you have the Statistics Toolbox is to use cvpartition. There are various ways to use it, from the simplest kind of "hold out" scheme that you describe, to more complicated k-fold cross-validation.


Find more on Statistics and Machine Learning Toolbox in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!