How can I bin Data

My problem is the following:

How can I bin the data in this case: I have a matrix with column 1 x with 90 values of price and column 2 y with 90 values of demand. I want to bin the x values for example

x<30 as x1, 30<x<35 as x2,35<x<40 as x3, 40<x<45 as x4,45<x<47 as x5,47<x49 as x6, 49<x<52 as x7 and x>52 as x8. And I want to have for each x1,x2,..,x8 all the replicates of y in a matrix.

Then to find the mean of y-s for each x1,x2,...,x8.

Answer 1

[Star Strider]

There are probalby several ways.

One approach:

x = randi(90, 90, 1);                                   % Create Data
y = rand(90,1) .* sin(2*pi*x/90);                       % Create Data

partitions = [0 30; 30 35; 35 40; 40 45; 45 47; 47 49; 49 52; 52 max(x)+1];

for k1 = 1:size(partitions,1)
    ypartmean(k1) = mean(y((x >= partitions(k1,1)) & (x < partitions(k1,2))));
end

Experiment to get the result you want.

[gjashta]

I need also to include the histogram bin counts.

[Star Strider]

Add one assignment:

for k1 = 1:size(partitions,1)
    ypartmean(k1) = mean(y((x >= partitions(k1,1)) & (x < partitions(k1,2))));
    binct(k1) = numel(y((x >= partitions(k1,1)) & (x < partitions(k1,2))));
end

The ‘binct’ vector will have the bin counts.

[gjashta]

Thank you! How can I see the histogram?

[gjashta]

Attached is how I am trying to bin the data.

[Star Strider]

My pleasure.

Try this:

for k1 = 1:size(partitions,1)
    ypartmean(k1) = mean(y((x >= partitions(k1,1)) & (x < partitions(k1,2))));
    binct(k1) = numel(y((x >= partitions(k1,1)) & (x < partitions(k1,2))));
end

ctrs = median(partitions,2);
figure
hb = bar(ctrs, binct);
xt = get(gca,'XTick');
xtl = sprintfc('%d-%d', partitions)';
set(gca, 'XTick',ctrs, 'XTickLabel',xtl', 'XTickLabelRotation',45)

This would have been much easier if you had provided your data, and mentioned all these additional requirements in your original post.

[gjashta]

Thanks a lot Star Strider! Sorry for confusing you!

[Star Strider]

No worries, and no confusion. It simply would have been easier to have addressed all this at once.

If my Answer helped you solve your problem, please Accept it!

Answer 2

[Steven Lord]

If I understand your application correctly use histcounts2 to bin the data. The fourth and fifth output are the indices of the bins into which each element of your x and y data is binned. Use that bin information as the grouping variables in a call to groupsummary. See the "Multiple Grouping Vectors for Vector Input" example on the groupsummary documentation page for an illustration of the latter part of this technique.