I've noticed that there is a Minimum Bounding Geometry tool now. How might I use this tool (or related tool within ArcGIS) to create a 95% Minimum Convex Polygon. It would appear that the before mentioned tool is performing a 100% MCP.
In home range analysis, our department has used 95% MCPs (Minimum Convex Polygons) to develop animal home range polygon locations. The "Minimum Bounding Geometry" tool in ArcGIS 10 is essentially a 100% MCP tool (mapping a polygon based on the outside point positions of a cluster). I would like to create a polygon that uses the 95% confidence clustering (which helps remove outliers) of points.
references to literature would help... some issues... creating the convex hull (CH) (aka convex polygon) and removing the hull points and re-doing the analysis will fail in almost all cases ... as will buffering negatively by 5% of the area not fairly represent your analysis, so in short could you provide some visuals or some description which would account for the general situations ie your clustering consists of 4 points, 5 points with one outlier etc etc. it may be the clustering algorithm that you are interested in rather than the CH
Hi, I am in the same situation as the original poster in that I too would like to construct 95% minimum convex polygons.
Mr. Patterson, I think you may be right in that this may be a clustering algorithm issue rather than the convex hull tool itself. The only example I can provide that better describes our question is this:
I have a distribution of animal movement points, i.e. lat/long (WGS 84), and I have just used the Minimum Bounding Geometry to construct a convex hull that encapsulates all of my points (see attached document). However, it is apparent that there are areas within my polygon that contain no points (empty) or are not near any other points, i.e. outliers. Essentially, this polygon represents 100% of my data points--this is what I believe the original poster meant by a 100% MCP. The downside of this method is that it produces a "home range", i.e. convex hull, that incorporates areas never used by the animals. Therefore, as a means to alleviate this downside, researchers disregard or remove the top 5% of the data points in hopes of reducing the impact of outliers or large data points. The new "home range," then, is 95% of the original 100%.
I am not aware of any existing tool that can give you a convex hull in terms of x%. But you can check out Aggregate Points tool (ArcGIS 10, ArcInfo, Cartography - Generalization), which creates polygons around points based on distance. Or you can use Buffer tool to create dissolved polygons around points based on buffer distance. In either case, you can delete any small polygons in the output by area or by point count covered by the polygon. Are the remaining polygon(s) good for your need? If not, you can use the remaining polygons to select Contained points; and use these points to create the convex hull using the MBG tool.