私はArcMapで、重症度を含む事故データのヒートマップを処理しようとしています。これはArcMapの処理に関する質問ではありません。密度レイヤーの生成方法は理解しています。<\/P>
事故の重症度に重みを付けて、実際の安全性の緊急性をよりよく表現するための標準的または受け入れられている方法論について知っている方や、参考になる資料を教えていただける方がいればと思っています。<\/P>
事前に感謝します。<\/P><\/BODY><\/HTML>
Jeff,
I appreciate this perspective and the resource.
I have run some heat maps based upon straight one to one density as well as on the 1-5 weighting and I am interested in doing something for comparison based on what you mentioned.
I have looked over the documentation link you provided and am trying to ascertain what the remaining KABCO weights might be numerically and basis for the quoted "230 time" factor for fatalities you mentioned.
Can you cite a specific table or page in the document?
Thanks in advance
Another resource is the Crash Analysis tools available here: Crash Analysis | ArcGIS for State Government.
These are based on the methodology defined by the US Road Assessment Program (usRAP). The result of this analysis produces the following maps: Crash Density, Crash Rate, Crash Rate Ratio, and Potential Crash Savings. This analysis is different from other kinds of "heat maps" in that it compares similar segments so the final potential crash savings map is a good indicator of where countermeasures should be applied. There is also the Crash Risks Briefing | ArcGIS for State Government that goes through all of the analysis and provides a good way to communicate your results.
I have used the economic cost of a crash given in Blincoe 2015 Table 1-4
https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812013
Yields vastly different results than the 1-5 scale. A fatal crash is weighted 230 times a PDO crash.
Building on what Scott has said, of course 1 through 5 doesn't tell the whole story. It's an attempt at simplicity for a massively complicated dataset. 54 columns of data seems a little small, but I'm sure it has nearly everything you need to hopefully have a summary for each point. There will be a column (or columns) mentioning safety belts and this could impact what may have happened whether or not it is worn.
In any case, I think it's a great application to work on and one that can potentially tell a lot of stories with the data you already have. I would suggest just trying to make some data correlate with each other based on the criteria you put in (and expand from there). Though, much of this will be up to the decision makers and who is needing to see this data. There will be many opinions on this and you may have to make something separate for each person that needs this data.
Let me add this group 911 GIS . I know they are on the front-end, but I suspect get inquiries from researchers in the aftermath and may be able to provide some insight.
Chris Donohue, GISP
<chuckle> and by the way...this is my first application of my college degree and in an intern capacity so, I am open to whatever directions any might wish to point me in.
Scott,
Good points that in the process will likely be explored and reported as we analyze the initial results. I do have all of that data from our state DOT that they can release for 2013-2014. (54 columns of details per incident row)
As a matter of perspective, this is all part of both local reporting and for the big target of 5 year reporting by the local Metro Planning to USDOT.
The initial results I have generated are promising and as always create as many questions as answers but, sometimes you just have to throw something with reasonable validity up in order to know what questions you wish to ask further.
So yes, I am looking at drilling down particularly for local action as to what mitigating factors may be involved but, all things being equal, we want to analyze corridors and intersections to see if there are any improvements that can be made there first as this is a matter of reporting for local transportation funding.
Appreciate the input.
Looks like Adrian has given great info for a rating scheme and I held off commenting to see what all was brought up. Reason I am commenting was to better understand the analysis DOT's put behind this and I saw the original post mentions "better represent the actual safety urgency".
Some of my first thoughts were is the intention to understand hot spots based on personal injury or property damager or both? Is it neither and it's trying to understand potential safety risk factors in an area that potentially correlate with the higher/severe incidents with hopes of being able to identify and apply mitigating factors against that risk? How do you account for false positives such as a 5-fatal that is due to not wearing a seatbelt but had a seatbelt been worn would have been a 1 or 2? Are there seasonal considerations or specific weather considerations accounted for?
May not even be the depth or clarity you are working towards which is fine and this is just more me wanting to understand the logic behind analysis for these things. Thanks
You can kind of think of the numbers as a coded domain that can be quantified
1 - no injury
2 - possible injury
3 - minor injury
4 - serious injury
5 - fatal
Todd,
Yes, it is multiplicative. It isn't exact but it's usually a good representation. (meaning a fatal crash isn't 5 times worse than a non-injury crash). Usually the police office writing the report will populate the data on the crash severity and then that data gets updated as the person moves along (like if it is fatal later, etc.).
Interpolating these points into a raster would be very interesting and could possibly show you greater severity areas (using density clusters and such). This would make for an interesting study granted that you have all the data that you need.
Adrian,
Thanks for the input and help. From your answer I am taking that this is being done as a straight arithmetic function then and multiplicative etc?
I am considering also doing a raster classification and then selecting by location to calculate out a probability of fatality, class A, class B etc for the various density clusters. Any thoughts on that?
I have seen it done on a 1 to 5 scale. 5 is fatal and 1 is very minor - enough to just be reported (no injury). 4 is typically critical condition, etc. That way, you can easily sort out the fatals and then see a weighted approach to crashes in your area.
サインインしたメンバーは投稿、更新のフォローなどができます。初めてですか？無料アカウントを登録してください。
Find useful guides, FAQs, and documents to help you navigate and make the most of Esri Community.
