I have 5 arrays (A,B,C,D,E) of size(20x1). 20 Rows represents 20 models. I want to rank best model based on minimise (A,B), maximise (C,D,E). Any hint will be helpful.

A,B,C,D,E

[Mathieu NOE]

have code ? data ?

[Avijit Paul]

No code. I just want to know how to proceed to the problem.

A= rand(20,1);

B=rand(20,1);

C=rand(20,1);

D=rand(20,1);

E=rand(20,1)

Select and rank the best possible row according the condition minimum(A,B) and maximum(C,D,E)

[Torsten]

Say you have the rows [A B C D E] as

[1 -5 1 1 9]

[-7 3 1 3 7]

Which is better and why ?

[Avijit Paul]

So accoding to problem statement best (-7 -5 1 3 9). I think, we can rank [ (2 1); (1 2); (1 1); (2 1);(1 2)] . And add the ranks to get final value. Then sort the final value to get the ranking. This can be possible solution. For this case (7 7). Solution is all with equal importance . This can be one way thinking. Thats why I was asking for any help.

Answer 1

[Mathieu NOE]

maybe this ?

AB= rand(20,2); 
CDE=rand(20,3);

ABinv = 1./AB; % minimum(A,B) replaced by maximum of it's inverse

ABinvCDE_max = max([ABinv CDE],[],2); % find max (rank) along direction 2 
[val_sorted,rk] = sort(ABinvCDE_max);% rank = rk