how to do a conditionnal mean

Hi all,

I have many variables in a dataset.

One of them is called : FYEAR and another one is called : DY

I would like to find the mean of DY for each FYEAR and plot them.

Is there anyway i could do this ? can i do it with a boucle (IF for example ) ?

Thanks.

[Matt J]

can i do it with a boucle

This kind of boucle?

http://en.wikipedia.org/wiki/Boucl%C3%A9

[Guillaume]

'boucle' is french for 'loop'

Answer 1

[the cyclist]

You can use accumarray() to do this:

FYEAR = [2001; 2001; 2002; 2002];
DY = [1; 2; 5; 6];

[uniqueYear,i,j] = unique(FYEAR);

meanDY = accumarray(j,DY,[],@mean)

uniqueYear is the list of years in your list, and meanDY is the mean values.

There are many way to plot them. Here is a simple one:

plot(uniqueYear,meanDY,'.-')

[alex monjambo]

Thanks to all :) !

Answer 2

[Jos (10584)]

Something along these lines should work (assuming FYEAR and DY are numerical arrays)

FYEAR = [10 10 11 12 11 10 12 12 10]
DY =    [ 1  3  4  8  6  2 10  9  2]
[UniqueFYEAR,~,i] = unique(FYEAR,'stable')
MeanDY = accumarray(k,DY,[],@mean)
plot(UniqueFYEAR, MeanDY, 'bo')

Answer 3

[Matt J]

Here is yet another accumarray-based approach, but avoids a call to unique(), and also uses a faster 2-pass calculation,

FYEAR = [10 10 11 12 11 10 12 12 10];
DY =    [ 1  3  4  8  6  2 10  9  2];

counts=accumarray(FYEAR(:),1)
sums=accumarray(FYEAR(:),DY(:));

idx=logical(counts);

means(idx)=sums(idx)./counts(idx);