How can I create in Matlab a matrix (20 X 12) with random distribution of numbers 1 and 0, when in each column I must have 40% of numbers 1 and 60% of numbers 0? this must be random distribution.
Anyone could help me?
Thanks a ton!
Add a comment
|
3 Answers
An efficient method is:
- Generate matrix of uniform random values between 0 and 1.
- For each column compute the 40-percentile.
- For each column, set to 1 the entries that are lower or equal than the computed percentile, and set to 0 the remaining entries. This assures the desired fraction of values per column.
This can be done easily with prctile and bsxfun:
rows = 20;
cols = 12;
p = 40; %// percent of 1 values
A = rand(rows,cols); %// uniform random values between 0 and 1
perc = prctile(A,p); %// percentile of each column
A = bsxfun(@le, A, perc); %// 1 if lower or equal than percentile, 0 otherwise
1 Comment
user3432905
perfect..this function is new for me..i will check out in help. thanks
Here is a robust method, set the first 40% of each column to be 1 and then just randomly reorder each column.
m=20;
n=12;
M = zeros(m,n);
M(1:round(m*0.4),:) = 1;
for col = 1:n
M(:,col) = M(randperm(m), col);
end
1 Comment
user3432905
thanks Dan. i was also using the same approach but was not getting correct output .finally found my mistake from ur reply.
This could be one "aggressive" approach -
N = 10000;%%// A big number to choose 12 columns from
A = round(rand(20,N));
out = A(:,find(sum(A,1)==round(0.4*size(A,1)),12))
Let us know if this works for you!
4 Comments
user3432905
thanks. it worked but what is the purpose of having large value of N? i didnt understand its use !!
Divakar
Basically we are creating a large matrix (20xN) and from that choosing the first 12 columns that satisfy the criteria of 40% ones. Pretty easy to understand and use, if you ask me :)
user3432905
yes i got..you are actually selecting 12 columns satisfying given condition.having limited N say 30 wouldn't make it sure that required number of entries will be obtained..
Divakar
Exactly, that's why "aggressive".