How to write a code for 3 Variables

I want to write a code to obtain a new value array that is dependent to 3 variables. My 3 variables are like the table shown below:

A                    B                         C
1                    400                       200
1                    500                       100 
2                    455                       300
2                    200                       400
3                    100                       500
3                    500                       600
2                    600                       100
3                    700                       900                        
1                    800                       150
2                    900                       150

I want to calculate new 'D' array from A,B,C. My code should be like this if A=1 then calculate the sum of B and divide it to sum of C.

For example we have 3 row that A=1 and we have B=400,500,900 C=200,100,150 for A=1. So it shoulde be 400 +500+900 /200+100+150. The result (D) is 4 for A=1. I have so many A values and ı don't know how to calculate all of them. If you help me, I will be very happy. Thank you.

Answer 1

[John D'Errico]

Easy peasy. What, 2 lines?

abc = [1    400    200
1     500     100 
2     455     300
2     200     400
3     100     500
3     500     600
2     600     100
3     700     900                        
1     800     150
2     900     150];

Download consolidator from the file exchange.

[ai,bcsum] = consolidator(abc(:,1),abc(:,2:3),@sum)
ai =
   1
   2
   3
bcsum =
      1700         450
      2155         950
      1300        2000

bcratio = bcsum(:,1)./bcsum(:,2)
bcratio =
     3.7778
     2.2684
       0.65

There is no need for the A values to be integers. As long as they are distinct will suffice.

[Gokhan Kayan]

Thank you very much John this is what I exactly want :)

Answer 2

[Walter Roberson]

sumB = accumarray(A, B);
sumC = accumarray(A, C);
D = sumB ./ sumC;
D(sumB == 0 & sumC == 0) = 0;

This applies directly only if the A values are positive integers, preferably small and consecutive. If they are not positive integers then there is an adjustment that can be made using unique()

The final setting to 0 is for the case where the sum of B and sum of C are both 0, replacing the NaN that would result with 0. The sums could be 0 if the entries can be positive and negative; the sums can also be 0 if there is a gap in the values of A, such as if A might be [1, 2, 4] with no 3 entry.

[Gokhan Kayan]

Thanks Walter :)

Answer 3

[Geoff Hayes]

Gokhan - you can do

 A==1

to return an array of logical values, zeros and ones, that will tell you which element of A is a one (indicated by a one) and which element of A is not a one (indicated by a zero). For example,

 A = [1 2 3 4 5 1 1]

then

 A==1

returns

 1     0     0     0     0     1     1

And so you could then do

 sum(B(A==1))

to sum those elements of B that correspond to ones within A. See logical indexing for more details.

[Gokhan Kayan]

Thanks for your respond Geoff :)