Select to view content in your preferred language

# Clustering Analysis for combined number of element group

192
4
3 weeks ago
Labels (1)
New Contributor II

Hi guys,

I cannot locate any reference to some kind of analysis that I'd like to perform.

For my sport club, I want to define a set of groups with a variable number of elements that can be of 10, 11 or 12 elements. I'm having a total of 141 clubs spread of the National territory which needs to be clustered into groups. I'm seeing quite hard to make these groups even, and to not have some of the elements too far from their peers from its group.

Locations: I have a set of 141 locations, each defined by coordinates (X, Y).

Groups: I want to cluster these locations into groups of size between 10 and 12.

Objective: My goal is to form groups that are as efficient as possible in terms of geographic proximity.

My question is:

Is there a model/tool that allows me to evaluate all possible group combinations given the size constraints and optimize for the proximity of locations within each group?

I have looked into the current ArcGIS Pro tools for clustering, but I see no way to apply for specific case.

Any guidance or references to methods, algorithms, or theories that can be applied to this problem would be greatly appreciated.

Attached the dataset containing a CSV with the X,Y position and ID.

4 Replies
by
Occasional Contributor II
New Contributor II

I've checked the different tools. What I was still need is to have the possibility to constraint the number of elements of each group, like [10-12] elements.
The tools that I've seen in Pro is only giving me constraint of the number of Clusters, but not including the number of features inside.

I'll continue my research. Thanks!

MVP Esteemed Contributor

I suspsect you are wanting to tile/subdivide your data on some attribute or geometry.

many of the clustering algorithms depend on number of clusters as you can see from the common examples in python solutions

2.3. Clustering — scikit-learn 1.5.1 documentation

... sort of retired...
New Contributor II

Thanks @Dan for the Clustering link.
My data is quite simple, X,Y point features that need to be gathered into groups of 10..12.
My intention is to have several compositions of these groups and also reflect the amount of distance between each member group versus its peers from same group. So, I can check if some of the elements should be re-group to not to have a large distance.

I'm still figuring out how to build that.

Thanks,

Juanma