# Comparing more than two columns of a matrix using '<' or '>' symbol

7 views (last 30 days)
pavikirthi on 26 Oct 2015
Commented: pavikirthi on 26 Oct 2015
How to compare more than two columns of a matrix using '<' or '>' symbol?
For example, if I have a matrix A and if I compare the columns using '<' symbol
A=
2 5 4
4 1 2
6 0 1
7 5 8
I should get the output as
1 0 0
0 1 0
0 1 0
0 1 0
So, is there any solution that doesn't restrict for any two columns or three columns but in general more than two columns

Guillaume on 26 Oct 2015
There is only ever one number per row that is smaller than all the other columns, and that is the minimum of the row. Hence, use min:
A = [2 5 4; 4 1 2; 6 0 1; 7 5 8];
[~, col] = min(A, [], 2);
B = zeros(size(A));
B(sub2ind(size(A), 1:size(A, 1), col')) = 1

#### 1 Comment

pavikirthi on 26 Oct 2015
Thanks, it absolutely works fine for any number of columns.

Steven Lord on 26 Oct 2015
Take a look at the second output of the MIN function.
A = [2 5 4; 4 1 2; 6 0 1; 7 5 8];
[~, index] = min(A, [], 2)
Then use either a FOR loop or SUB2IND to fill in the appropriate elements in your output matrix.

#### 1 Comment

Stephen Cobeldick on 26 Oct 2015
>> A = [2 5 4; 4 1 2; 6 0 1; 7 5 8];
>> [~,idx] = min(A,[],2);
>> X = false(size(A));
>> X(sub2ind(size(A),1:size(A,1),idx.')) = true
X =
1 0 0
0 1 0
0 1 0
0 1 0

Lessmann on 26 Oct 2015
Hi,
you can achieve this by using
B = diff(-A,1,2) < 0
You have to concatenate a column of zeros, so that A and B are of equal size.

#### 1 Comment

Thorsten on 26 Oct 2015
That's only valid if you want to restrict your comparison to adjacent columns. In the example, column 1 is not compared to 3.

Thorsten on 26 Oct 2015
A = [2 5 4
4 1 2
6 0 1
7 5 8];
pairs = nchoosek(1:size(A,2), 2)
X = cell2mat(arrayfun(@(i)( A(:, pairs(i,1)) < A(:,pairs(i,2))), 1:3, ...
'UniformOutput', false))

#### 1 Comment

pavikirthi on 26 Oct 2015
Thanks. But, I want to compare whether column 1 elements are '<' column 2 and column 3 and similarly column 2 elements are '<' column 1 and column 3 and lastly column 3 elements are '<' column 1 and column 2.
I want to perform this operation in general with any number of columns. So, for this example of 3 column matrix I should get the output as
1 0 0
0 1 0
0 1 0
0 1 0