Can I have a code that detects # of element transitions and return a matrix that has only rows of minimum element transitions?

I have A=[1 0 0 2;1 0 2 2;2 0 1 2;0 0 0 2];

I want a code that eliminates the rows that have multiple element changes compared to other rows:

so in A, first row and second row has only one element change 3rd element ((I call it element transition)) so the code will return a new matrix B that has only the first two rows;

B=[1 0 0 2;1 0 2 2]

Answer 1

[James Tursa]

Assuming you compare a row to the previous row and allow at most one element to change, you can use the results of diff(A), count the number of elements that have changed, and discard all rows that have more than 1 change. E.g., (assuming A is not empty)

B = A([true;~(sum(logical(diff(A,1,1)),2)>1)],:);

[JacobM]

Great!

Should we have a reference to compare to? I am thinking if MATLAB can find the minimum transitions among all, so it may be useful if I compare all rows to each other, i.e row 2, 3 , 4 are compared to 1 then row 3, 4 compared to 2 and so on. So by that I am certain that the returned row has minimum #transition of all elements compared to all rows, does this make sense? what do you think?

Answer 2

[Andrei Bobrov]

...row 2, 3 , 4 are compared to 1 then row 3, 4 compared to 2 and so on...

C = squeeze(sum(bsxfun(@minus,A,permute(A,[3,2,1]))~=0,2));
[ii,jj] = find(min(C(:)) == C);
B = arrayfun(@(x,y)A([x,y],:),jj,ii,'un',0);