Grouping points on scatter matrix

**JamesWM** · 02-10-2022, 01:59 PM

Hello All,

Wondering if someone of you might please be able to help?

If you consider a full matrix of data as below but if it continued to cover a grid of 100 by 100

(1;3) (2;3) (3:3)
(1:2) (2,1) (2;2)
(1;1) (2;1) (3;1)

If I have the Weighted average of all the points in that matric and then plot that average in the matrix itself - I'm trying to know if its possible (and how) to group all points/items etc. that call on either side of the slopes... Hopefully the image attached helps explain. (???)

Matrix data.JPG

Each of the tiny blue data points are coordinates of the matrix - and since its on a standard Cartesian plane... I have the slopes etc. I have been trying to group the coordinates that would fall within the area of A and B based on slopes and angles of intersection etc. - but there are instances where the process I've developed labels coordinates that fall into area B - under the naming of Area A.

Any ideas of was to solves this would tremendous as I've been at this for what seems like a 100 man hours now and still keep running into issues.

A final mention - my goal is to achieve this process on a 2x2 grid (so 4 quadrants and each quadrant is further divided by the lines of the quadrants weighted average (visually shown as the larger blue circles in the images). Each coordinate in the grid matrix is at an internal of 0.01. Mentioning this, as the top right quadrant has negative slopes that I've struggled to work around.

XY scatter matrix.JPG

Any help - would be ever so grateful!

Thank you for the time and for even reading and trying to understand my explanation.

Thanks,

**MrShorty** · 02-10-2022, 02:18 PM

I have been trying to group the coordinates that would fall within the area of A and B based on slopes and angles of intersection etc. - but there are instances where the process I've developed labels coordinates that fall into area B - under the naming of Area A.

It sound like you already have an algorithm in mind that you are trying to use, but it is missing a few points.

Do you want us to help debug why the current algorithm is failing on a few points? If so, then we probably need to know the algorithm you are currently using, examples of points where the algorithm works and examples of points where it fails, and how you are determining that the algorithm is failing.

Or maybe you want us to come up with a brand new algorithm? In that case, we probably need to know enough about your current algorithm so that we don't just come up with the same algorithm.

I guess in either case, we need to know something about your current algorithm -- either so we can help debug it or so we don't duplicate it.

I would also suggest that it would dramatically help us help you to upload a sample file with a good, representative sample of data points. It probably needs to include points that obviously fit in A or B as well as points near the boundary where it is more difficult to determine which category to assign so that we can test either your current algorithm or our newly developed algorithms to see how well they work on a sample data set.