I have a data array:
A B C D
1 75 84 96 45
2 56 99 74 32
3 32 93 56 55
Which I rank using the RANK function:
A B C D
4 3 1 3 2
5 2 3 2 1
6 1 2 1 3
I then set-up a 3rd array of weighted values:
A B C D E F
7 0 0 0 0 sum=(A7:D7) =E7/$E$10
8 0 0 0 0 sum=(A8:D8) =E8/$E$10
9 0 0 0 0 sum=(A9:D9) =E9/$E$10
10 sum=(E7:E9)
which is computed by 4 variables such that given:
A B C D
10 x1 x2 x3 x4
then:
A7 = A4 * $A$10
B7 = B4 * $B$10 and so forth
I then sum the rows of the weighted array and calculate the % of total row sum.
Finally, I use Solver to calculate (Evolutionary) a max weighted row % in any order I define.
Specific example: Let's say I want to solve for F8 > F9 > F7
I would set up Solver to max F8
using variables (A10:D10)
subject to these constraints:
all variables >= 0
all variables <= 10
sum of all variables = 10
F8 = LARGE(F7:F9,1)
F9 = LARGE(F7:F9,2)
F7 = LARGE(F7:F9,3)
This system works great with simple arrays. Can anyone suggest a method that is FAST on very large arrays with many variables to compute?
Thanks!
Bookmarks