Grouping 2d data with different bins for each dimension.

Dear guys I'm trying to group my 2d data but with poor results.

My input are x and y. At first I divide them into bins with different dimensions as it follows:

a=(min(x):0.2:max(x))'; %17x1
b=(min(y):0.2:max(y))'; %34x1

At this point through I asses the distributions in each bin:

N=histcounts2(x,y,a,b);

Finally, I subplot my data in order to have at the left side the scatter of points while at the other an heatmap:

subplot(1, 2, 1);
scatter(x, y, 'r.');
subplot(1, 2, 2);
imagesc(a, b, N);

Unfortunately this is my result

The right side is correct but somehow in the wrong side. How can I correct this?

Answer 1

[Cris LaPierre]

An image axis has the origin and the top left while a cartesian axis has the origin at the bottom left.

Counsider using histogram2 with the 'DisplayStyle' set to 'tile' instead of histcounts2 and imagesc. See this example.

[Riccardo Tronconi]

@Cris LaPierre thank you very much. Just an extra question. For each bin instead of display the absolute number, would it be possible using relative numbers?

Lets say the bin 10 has =50 values but the total values recorded are 100 so the relative number related to that bin is 50/100= 0.5. Is there a way to do that?

[Cris LaPierre]

I'm not sure if that specific option exists, but look into the Normalization options. This is one of the name-value pair arguments for histogram2. Options are

• 'count' (default)
• 'probability'
• 'countdensity'
• 'pdf'
• 'cumcount'
• 'cdf'