Multiple Gravity Centers

832
5
08-21-2017 06:48 PM
JavadRouzafzoon
New Contributor II

Hello,

I have customer information as a map layer, and I would like to set and find the number of required gravity centers for building collection points.

Central Feature and Mean Center tools work fine, but the only problem is that I can not define the number of gravity centers needed.

Is there any other tool or method which I can use?

Thank you!

ArcGIS Educational License

0 Kudos
5 Replies
DanPatterson_Retired
MVP Emeritus

There are options, but it depends on your point pattern structure.  On of the simplest is to use the locations for form the clusters (as polygons), then take the centroid of that to form the centre.  You specify the distance rather than the number of clusters.

See Aggregate Points ... 

Grouping Analysis ... allows for attribute grouping with a spatial addition and maybe

Cartographic Partitions ... and there are more clustering algorithms (see Dr Wiki for a generic overview ... a lot of which are implemented in some form in Arc* software)

JavadRouzafzoon
New Contributor II

Hi Dan,

I have 100,000 points with different factor weights. Can these tools process this information?

what is the best approach or tool based on my data size?

Thank you for your time

0 Kudos
JaySandhu
Esri Regular Contributor

You can also take a look at Location-Allocation. It is designed to "locate" facilities that satisfy some distance or capacity constraints. One of the solver types is Minimize Facilities that will find the minimum number needed for a given distance cutoff. You need a street network dataset to use it with Network Analyst extension. Or you can use it in the Ready-To-Use services from ArcGIS.com

More details here:

Location-allocation analysis—Help | ArcGIS Desktop 

Jay Sandhu

JavadRouzafzoon
New Contributor II

Hi Jay,

Many thanks for your guidance.

I have 100,000 customer points information with different factor weights. Can I insert them both as demand points and candidates facilities? Because centers can be built on any of these points.

Thanks for your time.

0 Kudos
JaySandhu
Esri Regular Contributor

With 100,000 points as both demand and facility, and no spatial (distance/time) constraints, the process will likely run out of memory trying to compute/store the OD matrix with ten billion entries. I would suggest that you intelligently cut down on your facilities so that not all 100,000 are used. One way is to randomly select a smaller set, say 1000 facilities, using the "Create Random Points" GP tool. BTW, if you gave it 100,000 facilities and had a reasonable distance cutoff so that a full matrix is not computed, the 100,000 choose a few facilities is still a very large number of combinations to evaluate and get an answer in reasonable time. I strongly suggest not having more than a 1000 facilities with 100000 demand points, and still do add a distance cutoff.

Jay Sandhu