Original User: firefry82
Hi, first of all I want to say that I already have done a lot of reading about this, but I still think I'm missing something obvious.
I have point data, about 1200 points distributed over a large area (approx. 20,000 square kilometres). They are clearly clustered. You can see it with the naked eye and average nearest neighbor says so too. Now what I want to do is, I want to determine areas which contain a significantly higher density of points. These areas shall act as my study areas. So what I tried was this
:
1) ran an "average nearest neighbor" on my original data
[ATTACH=CONFIG]26947[/ATTACH]
2) built a model which first copies, than integrates the original data then collects events. I used the outcome (423,594424 meters) as tolerance for the integrate-tool.
3)ran a new "average nearest neighbor" on the result from my model (point 2)
4)startet "Incremental Spatial Autocorelation" (Spatial Statistics) (Morans I) with 1000 meter steps (increments) starting at 2877,357147 (result from point 3).
this is the result:
[ATTACH=CONFIG]26948[/ATTACH]
so the first peak is at 18877,36 meters (zscore=1,929998 und pvalue 0,05)
5) Now I ran "point density tool" (and kernel density, too) with 18877,36 meters as radius.
So far so good. But here comes my problem:
Depending on what classification method I use, the outcome differs greatly!!
Remember I wanted to use the most dens areas to determine my study areas. If the outcome differs so greatly I cannot do that I guess. Which classification method should I use? The data is not normal distributed and not linear. I tried standard deviation and quantiles, but I can still influence the outcome very much. I need some kind of statstical tool or formula that tells me where there is a clear break in the density data. Can anybody help me with that?
I'm using arcgis 10.1 with all licenses
this is what my point density looks like with different classification methods:
[ATTACH=CONFIG]26949[/ATTACH]
and they all still contain the zero values. If I cut them away (by excluding them), which I am not the least bit sure if I should do or not, then it all changes again...
[ATTACH=CONFIG]26950[/ATTACH]
