6

I have a data set of points which are grouped into clusters like so:

enter image description here

I'd like to use ArcGIS or Python to create a point layer of the centroids of each cluster but I can't seem to find the right tools. I looked into hot spot analysis but it doesn't seem to apply, since my features aren't numerically weighted. I also tried the Integrate tool with a reasonable tolerance but I didn't see any changes.

Improve this question

2 Answers 2

Reset to default
3

ArcGIS has a few tools in the Spatial Statistics toolbox that might be useful.

Mean Center

Identifies the geographic center (or the center of concentration) for a set of features.

Median Center

Identifies the location that minimizes overall Euclidean distance to the features in a dataset.

Central Feature

Identifies the most centrally located feature in a point, line, or polygon feature class.

I think the Mean Center would be most useful here when used with the Case Field option.

Improve this answer
3
  • I see now that previous comments describe what I just posted at approximately the same time. Commented Jun 11, 2014 at 0:18
  • The median / mean centre still needs the case field or it will create only one point. Commented Jun 11, 2014 at 0:21
  • Yes, that's true, and I mention that at the end of my answer. Commented Jun 11, 2014 at 0:26
2

To aggregate the clusters use grouping which produces a field SS_GROUP then for each unique SS_GROUP in the feature class create a convex hull then use Feature to Point to obtain the centroid of that particular cluster, with polygons you can specify inside to guarantee that banana shaped polygons produce a centriod that is inside the polygon but that wont matter too much as the convex hull cannot be banana shaped - that would be concave.

Improve this answer
9
  • Thanks for the tip - can I use Feature to Point with points as an input? I don't have polygons at the moment (those red circles are just for illustration). Commented Jun 10, 2014 at 23:40
  • Oh, I thought they were real. You want to cluster them and then get the centroid of that. Commented Jun 10, 2014 at 23:41
  • See edit. After grouping you may want to dissolve by SS_GROUP to a multipoint so it's one feature per group. I think that you can still create a convex hull over 1 multipoint feature. This would make iteration easier. Commented Jun 10, 2014 at 23:50
  • 1
    The Mean Center tool might also be useful here? I haven't used it but it looks like it's built for this purpose Commented Jun 11, 2014 at 0:12
  • 1
    I figured you could use Mean Center in conjunction with your SS_GROUP idea, using the case field to separate them into clusters (this may not give any advantages over the Feature to Point tool though) Commented Jun 11, 2014 at 0:15

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.