Select random data from a matrix and replace it

2 views (last 30 days)
Hello,
I have the following problem: I have a matrix e.g
1 0 1 0 1
1 1 1 0 0
1 1 1 1 1
1 1 0 0 1
and I try to do this: When the sum of each column exceeds the threshold=4 some random "1" of the column that goes beyond the threshold become zero. How do I implement this in matlabThis can be done without using loops?I would be very grateful if someone helpend me

Accepted Answer

Andrei Bobrov
Andrei Bobrov on 14 May 2019
Edited: Andrei Bobrov on 15 May 2019
A = [1 0 1 0 1
1 1 1 0 0
0 0 0 0 0
1 1 1 1 1
1 1 0 0 1
0 0 0 0 0];
p = 4;
[ii,jj] = find(A);
jjj = accumarray(ii,jj,[size(A,1),1],@(x){x(randperm(numel(x),min(numel(x),p)))});
k = cellfun(@numel,jjj);
out = accumarray([repelem((1:numel(k))',k),cell2mat(jjj)],1,size(A));
  5 Comments

Sign in to comment.

More Answers (2)

Jos (10584)
Jos (10584) on 15 May 2019
Edited: Jos (10584) on 15 May 2019
Here is another, indexing, approach:
A = randi(2, 6, 8)-1 % random 0/1 array
M = 3 % max number of 1's per column
szA = size(A) ;
B = zeros(szA) ;
tf = A == 0 ;
[~, P] = sort(rand(szA)) ; % randperm for matrix along columns
P(tf) = 0 ;
[~, r] = maxk(P, M) ; % rows with M highest values in P (including 0's) per column
i = r + szA(1)*(0:szA(2)-1) ; % convert to linear indices
B(i) = 1 ; % B contains M 1's per column
B(tf) = 0 % reset those that were 0 in A -> only maximally M 1's per column remain
  2 Comments
stelios loizidis
stelios loizidis on 16 May 2019
Works what I suggest and thank you very much!!!
I just have the following question: Once matrix B is created with
0 's and 1 's how it stays stable without changing put 0 's and 1's every time i run the code?
Jos (10584)
Jos (10584) on 16 May 2019
I don't get it. Why not simply run the code only once for a given matrix A?

Sign in to comment.


Jan
Jan on 16 May 2019
Edited: Jan on 16 May 2019
A logical mask is much simpler than handling the indices:
A = randi([0,1], 8, 8); % Test data
p = 4;
mask = (cumsum(A, 1) > p) & A;
A(mask) = randi([0,1], nnz(mask), 1);
This replaces 1s in each column by a 0 with a probability of 50%, but leaves the first p ones untouched.
But if you want to change any 1s without keeping the first p 1s of each column:
mask = (sum(A, 1) > p) & A; % cumsum -> sum, auto-expand: >= R2016b
A(mask) = randi([0,1], nnz(mask), 1);
If you want to replace the 1s in the columns with more than p 1s with another probability:
A(mask) = rand(nnz(mask), 1) > 0.85;
  3 Comments
Jan
Jan on 16 May 2019
Edited: Jan on 16 May 2019
@Jos: A maximum number of ones? Do you mean a minimum number of ones? I admit, I'm more and more confused.
Jos (10584)
Jos (10584) on 16 May 2019
From the OP: "When the sum of each column exceeds the threshold=4 some random "1" of the column that goes beyond the threshold become zero."
If I now read it again, just a single one in such a column should be set to 0 then? ...

Sign in to comment.

Categories

Find more on Numeric Types in Help Center and File Exchange

Tags

Community Treasure Hunt

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

Start Hunting!