I'm looking to create clusters both on spatial density, as well as an attribute (time, in seconds). Essentially, I am trying to figure out what areas are choke points in terms of time spent at certain intersections based on GPS data. So I've got XYs, as well as time in seconds spent at each location. The GPS points are taken at random intervals so I can't just query out the information I want beforehand.

Although creating clusters on spatial density works for 90% of my data, there are some instances where a GPS was stopped at a set of lights for a long time, but only happened a couple of times. I also want to identify those locations as well.

I seek any advice on either the analysis methodology, or specifics on how to cluster based on location and attribute.

  • I'm no expert in cluster analysis, but maybe you pre-select your points based upon their time and it is that selection you feed into your cluster tool? – Hornbydd Oct 10 '19 at 12:29

The Esri tool formerly known as the Grouping Analysis tool has evolved into two enhanced cluster analysis analysis tools that might apply to your problem:

Multivariate clustering will examine clusters solely based on attributes. (I think this is what you're looking for.)

Spatially constrained Multivariate Clustering examines for spatially contiguous clusters.

Not the answer you're looking for? Browse other questions tagged or ask your own question.