Random split of a vector of unequal lengths

How can I split a vector in k unequal subsets? For example, if I have 200 data, a random split might give us 63, 95 and 150. That is 1:63, 64:95, 96:150 and 151:200.

Apology for cross postings.

Answer 1

[Azzi Abdelmalek]

k=4
v=1:200
m=numel(v);
idx=unique([randperm(m-2,k)+1 m]);
idx0=[1 idx(1:end-1)+1];
out=arrayfun(@(ii,jj) v(ii:jj),idx0,idx,'un',0);
celldisp(out)

[Adib Zaman]

Thanks!

Answer 2

[Star Strider]

One way to do it:

A = 1:200;
ndiv = 3;       % NUMBER OF SUBMATRICES
idx = sort([1 randperm(length(A)-2, ndiv-1)+1 length(A)+1])
for k1 = 1:length(idx)-1
    R{k1} = A(idx(k1):idx(k1+1)-1);
end

The logic guarantees that every element of R has at least two elements. Choose the number of sub-matrices with ndiv.

Answer 3

[John D'Errico]

I don't see what is wrong with simply choosing k-1 values randomly between the min and max. Those points define a partition as you desire.

[Adib Zaman]

yes, the only problem is you need to sort it after the random number generation.