To determine the number of distinct pairs of value(s) in each column of matrix.

Given that i have a 4 by N matrix, each entry of the matrix can take a random value from 1-6,

A = randi([1,6],[N 4]) 

I want to determine how many column(s) of the matrix have 2 distinct pairs of any two numbers from 1-6, for example [1;5;5;1] or [6;6;1;1]

First off, i want to start simply by checking say if the first column of A has 2 '1's and 2 '2's so we want to see if the column is the same as any of { [1;1;2;2;] , [1;2;1;2] , [1;2;2;1] .... }

I think you can do it like this:

valCount = hist( B(:,1) , uniqueVals )'==[2;2;0;0;0;0]

Then i want to generalise this so it can check every possible pairs of values and spit out the number of columns with two distinct pairs of numbers, but i have not a clue how to proceed.

Please if someone could kindly help

Answer 1

[Turlough Hughes]

You could do the following. First find rows where there is only one other element equal to the element in the first column

idx = find(sum( A(:,1)==A,2 )==2);

Then make a new array B from rows satisfying the first condition and separate two values that didnt equal the first column. Then find indices where the remaining pairs are equal:

B = A(idx,:).'; % temporarily transposing to facilitate the reshaping step
remPairs = reshape(B(B~=B(1,:)),2,[]).';
idx2 = remPairs(:,1)==remPairs(:,2); % index where remaining pairs are equal
A(idx(idx2),:)