Counting specific element in one column corresponding to unique elements in an another column

I have two vectors a and b. I want to count how many times the value '-1' appears in b corresponding to unique elements in a. Is there an efficient way to do this without loops?

a = [1,1,1,1,1,2,2,2,2,2,3,3,4,4,4,5,5,5,5,5,5]';
b = [1,1,-1,1,-1,-1,-1,-1,1,1,-1,1,1,-1,-1,1,1,-1,-1,1,1]';

desired output:

1--2
2--3
3--1
4--2
5--2

Answer 1

[Jakob B. Nielsen]

You can use logic indexing. Example:

length(b(a==1 & b==-1));

You get those entries from b which fulfils the condition that the corresponding index position in a must equal 1, and also that the same index position in b must equal -1.

& is logic AND - both conditions must be true to get an output. If you need a logic OR, the matlab operator is |

[Hari krishnan]

Thanks for the reply. But i have a long vector for 'a'. Is there a way for giving values other than using loops?

[Jakob B. Nielsen]

Other than manual work, you probably wont get out of a loop entirely. But it is a very simple one;

%where N is your highest integer of a that you want the comparison for;
N=5;
for i=1:N
    output(i)=length(b(a==i & b==-1));
end

[Stephen23]

"Other than manual work, you probably wont get out of a loop entirely"

Basic MATLAB functions like histc don't require a loop.

Answer 2

[Stephen23]

>> a = [1;1;1;1;1;2;2;2;2;2;3;3;4;4;4;5;5;5;5;5;5];
>> b = [1;1;-1;1;-1;-1;-1;-1;1;1;-1;1;1;-1;-1;1;1;-1;-1;1;1];
>> u = unique(a);
>> n = histc(a(b==-1),u);
>> m = [u,n]
m =
     1     2
     2     3
     3     1
     4     2
     5     2