How can I partition the data to training (70%) testing (20%) & validation (10%) such that mean & SD of each subsets same?

I am trying to partition my database to training, testing and validation. However, the commands 'cvpartition' allows only training and testing. Also the mean and SD of partitioned data is not consistent.

Answer 1

[Pratham Shah]

Hello Srinivas!

You haven't mentioned with what kind of database you are working with. However if the data is related to image, you can try using 'spliteachlabel' function. Also to get 3 partitions instead of two you can follow this:

Make one partition of (80%-20%) and then you can divide the partition with 80% with (90%-10%) ratio.

Hope this helps :)

[Srinivas Alagesan]

Hi Pratham,

Thanks for your response. But I am dealing with numeric database having 11 features (input) and one output.