Randomly select samples from a matrix in proper order

4 views (last 30 days)
Shuo Zhang on 25 Jan 2018
Commented: Shuo Zhang on 26 Jan 2018
I have a 1 by 30000 matrix and I'd like to randomly select 300 elements from it. However, the selection should follow such a rule that if the previous selected element is the nth element in the matrix, then the next selection should be at least the n+1th element in the matrix. For example, if the first selection is the 5th element in the matrix, then the next selection should be done from the 6th element to the 30000th element, etc. I found out there's a command 'y = datasample(data,k)' that can take samples randomly, but what else should I do to ensure the right order?

Youssef Khmou on 26 Jan 2018
The selection procedure can be performed by ordering the inputs from the smallest to the largets, here is an example, a vector of N=4000 samples, we select randomnly P=100 points where the described rule is valid:
N=4000;
P=100;
x=rand(1,N);
Index=sort(round(N*rand(1,P)));
figure;
bar(Index);
y=x(Index);

James Tursa on 26 Jan 2018
Edited: James Tursa on 26 Jan 2018
Solution without repeats:
n = number of samples to select
y = x(sort(randperm(numel(x),n)));
If you have an earlier version of MATLAB, then that last part must be done in multiple steps:
r = randperm(numel(x));
y = x(sort(r(1:n)));

1 Comment

Shuo Zhang on 26 Jan 2018
This is quite amazing! Thank you very much!