How do I select every nth number of trials and put them in a separate column?

I have 180x4 matrix that is sorted by two columns. There are 5 levels or variable A and 6 levels of variable B (30 groups total). Variables C and D are just responses. I would like to sample from each group with replacement. E.g. This would mean that if had a loop with 30 iterations, I would select a sample from each of the 30 groups. If I had 60 iterations, I would select 2 samples from each group.

Answer 1

[KSSV]

For a vector

X = rand(100,1) ;
X10 = X(1:10:end) ;   % pick every 10'th value
X7 = X(1:7:end) ; % pick every 7'th value

For a matrix

X = rand(180,4) ;
X10 = X(1:10:end,:) ;   % pick every 10'th row
X9 = X(1:9:10,:) ; % pick every 9th row

[T Shep]

I guess my question could have been phrased better but this definitely helps!

Lets say that you have: X = rand(180,4)

And you would like to select a random sample from row 1-6. Another random sample from 7-12. A third random sample from the rows 13-18. A fourth random sample from rows 19-24 etc.......all the way to 175-180. Do you have any idea how to do that?

By random sample, I mean selecting that entire row..

[KSSV]

row = 1:180 ;
sample = reshape(row,6,[]);
pick = zeros(size(sample,2),1) ;
for i = 1:size(sample,2)
    pick(i) = randsample(sample(:,i),1) ;
end

[T Shep]

Thanks a lot!! This is perfect!!

[Jerod Hollen]

OH MY GOD THANK YOU

Answer 2

[per isakson]

Try this

>> [ B, S ] = cssm1()
B = 
    [ 6x2 double]    [3x2 double]    [6x2 double]    [3x2 double]    [11x2 double]    [9x2 double]
    [ 8x2 double]    [6x2 double]    [9x2 double]    [6x2 double]    [10x2 double]    [5x2 double]
    [10x2 double]    [2x2 double]    [3x2 double]    [9x2 double]    [ 5x2 double]    [4x2 double]
    [ 8x2 double]    [5x2 double]    [5x2 double]    [3x2 double]    [ 8x2 double]    [7x2 double]
    [ 8x2 double]    [4x2 double]    [4x2 double]    [5x2 double]    [ 3x2 double]    [5x2 double]
S = 
    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]
    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]
    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]
    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]
    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]    [1x2 double]

where

function    [ B, S ] = cssm1()

Create the 180x4 matrix, M

len = 180;
M   = nan( len, 4 );
M( :, 1   ) = randi( 5, len, 1 );
M( :, 2   ) = randi( 6, len, 1 );
M( :, 3:4 ) = randn(    len, 2 );

Binning

[ N, ~, ~, binX, binY ] = histcounts2( M(:,1), M(:,2), 'BinMethod','integers' );

Populate a bin container, B

B = cell( size(N) );
for ii = 1:5
   for jj = 1:6
       B{ii,jj} = M( binX==ii & binY==jj, 3:4 );
   end
end

Draw one random sample from each group

S = cell( size(B) ); 
for ii = 1:5
   for jj = 1:6
       ix = randi( size( B{ii,jj}, 1 ), 1 );
       S{ii,jj} = B{ii,jj}( ix, : );
   end
end
end

Notes:

• "question could have been phrased better" &nbsp I guess this is not a solution to your problem. This is rather an answer to the question in the Body than to that in the Title (subject line).
• "Lets say that you have: X = rand(180,4)" &nbsp with floats in column 1 and 2 'BinMethod','integers' need to be changed.
• histcounts2 was introduced in R2015b
• with a 180x4 matrix this function is fast enough(?), despite cell arrays and loops
• "selecting that entire row" &nbsp replace B{ii,jj} = M( binX==ii & binY==jj, 3:4 ); &nbsp by &nbsp B{ii,jj} = M( binX==ii & binY==jj, : );
• Finally: see Tables, which support features for binning and more.