Original User: hotspots
Hi everyone,
We are working on a Hot Spot Analysis for fatal car crashes in a region in New Zealand, using data from 2000-2012. We're undergraduate students who have been learning about ArcGIS across 3 years. However, we've never really done Spatial Stats so this is new to us.
What we've done so far is we used the Collect Events tool to create weighted points for the point data we have. Some of the events have been 'collected' into one point as expected.
We then ran the Hot Spot Analysis using Inverse Distance Band as the spatial relationship. This gave us really isolated hotspots and odd P-Values (just zero, and then all of a sudden jumping to 0.8)
We then re-ran the tool using Fixed Distance Band and it gave more logical results, but we had some outliers that were deemed statsitcally significant but were literally just a point surrounded by no other points.
Are we doing something wrong? Should we specify a figure in the "Distance Band or Threshold Distance" field for the Fixed Distance Band? We've read a lot of literature but can't seem to get our heads around it totally.
Any advice/comments would be appreciated.