traffic accident analysis - KDE in network space

Question

Hello guys,
I have to analyse traffic accident data in order to investigate the density of such point events. A standard (planar) Kernel Density Estimation (KDE) aims to produce a smooth density surface of spatial point events over a 2-D geographic space. However, the planar KDE may not be suited for analysis point events, which usually occur inside a linear space.

Does anyone have suggestion how to analyze traffic acciden data and detect hot spot using ArcGIS tools?

Thanks!

LaurenGriffin · Answer

Check out this story map and case study (complete with data and a detailed tutorial).  I hope it helps!

Lauren

KevinBell · Answer

Hotspot maps often don't tell the whole story (www.xkcd.com/1138)

Having said that, transportation engineers normalize the accident frequency in an intersection by the number of vehicles entering the intersection.  See the attached pdf (if you dare) This takes a few minutes to comprehend.

A normalize intersection rate can be obtained like so:

rate = (# crashes per yr * 1,000,000)/(# entering cars * 365)

This will give your the number of crashes per million entering vehicles.

2005-04-04_AccRatePMEVyearlyChange_Legend.pdf

sheet 2.jpg

JoeBorgione · Answer

Hello guys,
I have to analyse traffic accident data in order to investigate the density of such point events. A standard (planar) Kernel Density Estimation (KDE) aims to produce a smooth density surface of spatial point events over a 2-D geographic space. However, the planar KDE may not be suited for analysis point events, which usually occur inside a linear space.

Does anyone have suggestion how to analyze traffic acciden data and detect hot spot using ArcGIS tools?

Thanks!

And linear space isn't 2-d?

I guess you could perform a spatial join or a nearest analysis where your streets could assume the values of your point attributes.  You might need to do a few summaries here and there as well as a relationhip class or two, but you should be able map the street segments by numbers of accidents for a given segment.