partition column randomly in to three columns

I have a column consisting of (10000) rows that I want to randomly partition it into three columns, but as you can see, (10000) doesn't divided (3) so I couldn't use the ready-made Matlab functions to partition, I was getting an error.

could you please help me.

Thank you in advance.

[DGM]

You'll have to decide how you want to handle it. You can either pad the result (e.g. with zeros or NaN), or you can create three column vectors of unequal length (e.g. in a cell array), or you can interpolate the input vector (longer or shorter) such that the result is integer-divisible by 3, at which point you could just reshape or partition it.

[norh hameed]

I can't make these choices since the column I have is a distribution and I can't change it.

[norh hameed]

I know that some values will remain in the column and will not be included in the columns division, but I want the remainder or neglect to be as few as possible, in addition to that I want the selection to be random for columns.

[Dyuman Joshi]

The nearest value to 10000 that is divisible by 3 is 9999. So leave any 1 random value, and convert the rest to a 3 column matrix using reshape.

[norh hameed]

The problem is that the column size represents the size of the sample, i.e. it may change, it is not fixed, so I need a code that works in general, and I do not need to change or delete different values manually every time I run the program

Answer 1

[Bruno Luong]

A=rand(20,1)

A = 20×1

0.1011 0.6040 0.1700 0.4430 0.6715 0.6486 0.2120 0.1498 0.5527 0.7879

reshape(A(randperm(end,end-mod(end,3))),[],3)

ans = 6×3

0.9549 0.6715 0.1700 0.5665 0.2120 0.6040 0.6486 0.1150 0.1516 0.1011 0.6474 0.9099 0.0074 0.8325 0.4430 0.5527 0.8559 0.7879

[norh hameed]

Thank you so much Bruno Luong

Answer 2

[Walter Roberson]

L = size(YourData, 1);
N = floor(L/3);
G = [1*ones(1,N), 2*ones(1,N), 3*ones(1,N), randperm(3,L-3*N)];
G = G(randperm(L)) ;
S1 = YourData(G==1, :);
S2 = YourData(G==2, :);
S3 = YourData(G==3, :);

Note that this selects randomly but in this particular implementation the selected columns will be in their original order within each group.

This code randomly selects which groups are slightly shorter if needed.

[norh hameed]

The third column was larger than the other two columns, and I want the three columns to be the same size. I mentioned that we can neglect some values in the original column in order to divide the column equally and randomly into three columns, but we should neglect as little as possible.

Answer 3

[Bruno Luong]

The three parts has "almost" equal number of elements

A=rand(10000,1); % dummy test data

G=splitapply(@(x){x},A,randi(3,size(A)))

G = 3×1 cell array
    {3341×1 double}
    {3357×1 double}
    {3302×1 double}

% randomly shuffle
N = length(A);
N = floor(N/3)*3; % xomment this if you prefer not equal-length partition but none discard
G=splitapply(@(x){x},A(randperm(end,N)),mod((1:N)',3)+1);
A3=cat(2,G{:})

A3 = 3333×3

0.9759 0.0862 0.3139 0.9268 0.2455 0.1357 0.3654 0.6463 0.1973 0.1864 0.7527 0.2257 0.4267 0.9289 0.3331 0.4444 0.4423 0.3335 0.3562 0.4052 0.4297 0.7195 0.7529 0.0198 0.4037 0.1580 0.4706 0.3445 0.0772 0.1285

Then you can mix both of the above example methods.

[norh hameed]

Again, the size of the columns is not equal as you notice the second column contains 3334×1

[Bruno Luong]

@norh hameed Who tells they must be equal in length?

[norh hameed]

I need the columns to be equal because I want to combine them into one matrix, so the columns should be equal.

[Bruno Luong]

Then I just EDIT the code