MATLAB: How to perform nonrepeating bootstrap resampling in MATLAB

Statistics and Machine Learning Toolbox

Resampling with bootstrap contains the possibility that repeated samples
might occur. I would like to sample a set to obtain a subset in which the elements of the subset do not contain repeated results.

Best Answer

A workaround is to generate a set of unique random indexes.
See the following code for an example:
N = 1000;      % Size of your total set 
Nsub = 5;      % Size of the subset
nu = 0;
while(length(nu) < Nsub)
x = round(.5 + (N - eps(N+.5))*rand(1,10*Nsub));
[b1, nu, n1] = unique(x, 'first'); 
end
indexi = nu(1:Nsub)   % Unique and random indexes for subset into
                                     % the larger set.
Related Question