I have point features that have as an attribute an asset count. For instance, point A may have 3 assets whereas point B may have 300 assets.
My goal is to have each point be in two categories: clustered or non-clustered. A clustered point means that, in a buffer with, say, a 5 mile radius, has 100 assets total between all points in that buffer. A non-clustered point is one that doesn't fall in that definition. The points could only be counted once. It is also advantageous to have more clustered than non-clustered points.
One solution I had to this problem was to buffer every point and then for every buffer count the number of assets. If the number of assets were below 100, I would call the point associated with that buffer (the center point) non-clustered. Otherwise, it would be clustered. Then, I would go through the clustered buffers and start with the buffer with the most assets. Every point within the buffer would be clustered. Then I'd keep going with the points that were not in any buffer previously checked and then go through that process again until there were no more buffers.
The problem with that process is that it automatically assumes that the center for the buffers is a point in the data. It may be better to have buffers centered around points not in the data to get more clustered points. Is there a better way to go about this problem?
Programming is not an issue for me. I plan on making this a geoprocessing service if possible, so it's going to use 10 SP5.