I would like to sum a 2 dim array group by value in second column

I have an nX2 array

0.1 5

0.3 5

0.34 2

0.4 3

0.1 5

-.9 1

-6 2

and want to sum the first column grouped by the second to create an nX2 array that looks like

-0.9 1

-5.66 2

.4 3

0.5 5

I am sure its simple but its giving me brain ache :-)

Many thanks

Answer 1

[Abolfazl Chaman Motlagh]

there are a lot of ways to do this, here's a simple solution. i would recommend you to understand every line and run it line by line. try writing this code vectorize, i mean without for loop :)

X = [rand(6,1),randi(4,6,1)]

X = 6×2

0.2612 2.0000 0.6470 3.0000 0.4749 4.0000 0.6048 3.0000 0.6482 1.0000 0.1296 2.0000

Y = unique(X(:,2));
Y(end,2)=0;
for i=1:size(X,1)
    ind = X(i,2)==Y(:,1);
    Y(ind,2) = Y(ind,2) + X(i,1);
end
Y = flip(Y,2)

Y = 4×2

0.6482 1.0000 0.3909 2.0000 1.2518 3.0000 0.4749 4.0000

[simon lumsdon]

Many thanks, it works perfectly. Wow, had rather hoped there was a single line solution to this type of problem, some form of sum, group by approch, hadnt expected to have to tackle it long hand. Not sure I have the skills to turn that into one line of code though?

[Abolfazl Chaman Motlagh]

Here's a one line solution :)

X = [rand(6,1),randi(4,6,1)]

X = 6×2

0.8407 1.0000 0.1779 4.0000 0.2698 1.0000 0.7353 4.0000 0.8105 1.0000 0.1059 3.0000

Y = [sum((X(:,2) == unique(X(:,2))') .* repmat(X(:,1),1,numel(unique(X(:,2)))))' , unique(X(:,2))]

Y = 3×2

1.9210 1.0000 0.1059 3.0000 0.9132 4.0000