Is it appropriate to run a Hot Spots analysis on linearly referenced features (I want to find deer-vehicle collision hotspots on a highway and so I have 100-m segmented road with the number of collisions as events). The tool allows it but I am not sure whether the underlying assumption of randomly located collisions is being applied to only the linear features rather than the entire space within the extent of the linear features. In other words, is the tool detecting clusters merely because the collisions can only be on the linear feature (ie they cannot occur off-road) or does the tool "know" that the events can only occur on the linear feature?
I like the results I am getting but not sure about the underlying assumptions.
Thanks for any insights!
Cilla
