Binning x and y

Hi all,

Assuming that I have the follwing data:

x=[0, 2, 2.4, 2.8, 3.4, 3.9, 4, 4,7, 5.3, 5.8, 6, 6.4......]; % unit of x is cm

y=[100, 50, 30, 25, 20, 16, 14, 13, 12, 11, 10.5, 10, ....]; % Intensity

My question is the following:

I can plot y vs. x but I Would like, for instance, to bin x with a constant bin width (constant distance). Let say bin width of 2 (constant distance 2 cm). Therefore I would like to obtain a new x called xn:

xn=[0, 2, 4, 6,..]

and yn should contain the sum of the data which are conatained in in each interval:

interval1=[0,2] contains 100+50=150

interval2=]2,4] contains 50 + 30 + 25 + 20 + 16 + 14 =155

interval3=]4,6] contains 56.5

yn=[150, 155, 56.5]

and then I would like to plot yn versu xn

your hepl is highly welcome....

cheers

Answer 1

[the cyclist]

You should be able to do this with the hist() or histc() command. You can specify your bin locations, and then get the index for which x values went into each bin. Then use that index to sum your y values.

[Adam]

Hi ,

Thanks for your quick reply: I did it in the following way:

[n,bin] = histc(x,linspace(min(x),max(x),200)); yn=full(mean(sparse(1:length(x),bin,y)))

but this gives the mean and not the sum. xn now is xn=1:200 ? ist correct?

cheers

[the cyclist]

The first part is basically right, but I'm not quite sure what you're doing in the definition of yn. I would do this:

% Same as your code, but I explicitly define the bin edges as a variable
binEdge = linspace(min(x),max(x),200);
[n,bin] = histc(x,binEdge);

% Sum up the values in each bin.
yn = accumarray(bin',y);

% Plot
figure
plot(binEdge,yn)

The accumarray function is a powerful one. It is effectively looping through all of the bin numbers in the bin variable, and summing the corresponding values of y.

I think you are close to your final solution.

[Adam]

Hi ,

Thanks. My data points are symmetric (Fig.1) but when I bin these data using the previous code I obtain another data which are not perfectly symmetric (Fig. 2). Do you know why ? I kindly appreciate your help.

[the cyclist]

I suppose it is either

• the way the bin edges are defined, they are not centered around zero, OR
• the way MATLAB treats values that are exactly on the bin edge, it consistently puts them to one side, leading to asymmetry.

[laura9510]

Jumping in years later but if I was looking to average within bins, is using accumarray possible in conjunction with another function?