Would be really cool if you could add two fields: longitude/latitude of vehicle involved in collisions. You could make some really interesting maps with these data. Make it easier to identify collision hot spots. Could also link this to time component to see how collision location trends change over time.
National Collision Database
National Collision Database (NCDB) – a database containing all police-reported motor vehicle collisions on public roads in Canada. Selected variables (data elements) relating to fatal and injury collisions for the collisions from 1999 to the most recent available data.
- Publisher - Current Organization Name: Transport Canada
- Licence: Open Government Licence - Canada
|Resource Name||Resource Type||Format||Language||Links|
|National Collision Database 1999 to 2014||Dataset||CSV||
|Data Dictionary NCDB||Dataset||DOC||
|National Collision Database 1999 to 2014||Dataset||ZIP||
|National Collision Database 2014||Dataset||CSV||
|National Collision Database 2014||Dataset||ZIP||
[The following comment is posted on behalf of Transport Canada.] A good comment. Unfortunately, in our current database, the longitude and latitude are not available. It is something we hope the jurisdictions (who supply the collision data) will eventually be able to send us in the future.
The file is HUGE. It will be good if you separate it by year or API that makes it searchable by specific value as most people would like to view data with regard to specific type :car type, certain age, certain year...etc
It would also be great, if longitude and latitude are not available, to get some sense of where these collisions took place - even by province. Especially since motor vehicle licensing is by province, this way you could compare laws and policies with changes in collision rates over time.
The follow response has been posted on behalf of Transport Canada : "It has been determined by the jurisdictions that provide this data that province/territory is among a number of variables that cannot be shared. As a result, the information that is published must remain at the national summary level.” Thank you, Data.gc.ca
What does each record in the dataset represent? Does it represent a car involved in the accident or a person involved? I mean if two cars collided, each carrying two people, will there be 4 records associated to that accident, or only two? I have ruled out the possibility that each record represents one collision.
Thank you for your comment. The following is a response from the data owner: The National Collision Database is a hierarchal dataset with three main segments containing collision-level, vehicle-level and person-level information. Each record represents a person involved. In the example provided, where 2 vehicles each have 2 persons, there would be 4 records.
more clarification is needed about the data elements. medical fatalities versus collision severity fatalities is not clearly defined
Thank you for your comment. The following is a response from the data owner : If we understand the comment correctly, our response is that the difference between C_SEV and P_ISEV can be found in the data dictionary provided to the portal (and copied below). For clarification purposes, please note that C_SEV is at the collision level and that P_ISEV is at the person level. If there was something else, please re-phrase the comment/question. C_SEV Code Description 1 Collision producing at least one fatality 2 Collision producing non-fatal injury U Unknown X Jurisdiction does not provide this data element P_ISEV Code Description 1 No Injury 2 Injury 3 Fatality Died immediately or within the time limit. N Data element is not applicable e.g. “dummy" person record created for parked cars U Unknown e.g. applies to runaway cars X Jurisdiction does not provide this data element
This table is at the person level. this data is really hard to work with. I would really like to see the creation of a Incident level ID and a vehicle level ID so I can see how many vehicles/persons were involves in a collision. The IDs in the dataset are merely counts of number of people and vehicles.
Thank you for your comment. The following is a response from the data owner : Thank you for your comments. We will look into the feasibility of incorporating collision and vehicle level IDs into future iterations of this data product. In the meantime, please feel free to visit our online data tool at http://www.tc.gc.ca/VehicleCollisions. You could also contact Transport Canada’s Evaluation and Data Systems team directly via firstname.lastname@example.org
294MB file size? Surely there's a better way to present this data.
To my understand from the previous reply, there are multiple rows representing same incidents(one collision will possible generate multiple rows due to multiple injuries. Is there a way to identify which multiple row represents the same incidents so i can summarize them into one?
Fantastic Dataset. I echo the wish for Jurisdiction or provincial information but understand if that needs to be limited. Do you know if this will be updated past 2012? Thank you!
Hi Ray, the response from Transport Canada: "Thank you for your inquiry. At this time, the data are limited to national-level only. We are in the process of compiling the 2013 data so we expect to make these data available in the near future. Thank you."
I notice you state Bilingual in the various formats of the dataset, but don't see any French values. Do you have a list of all elements with French values?
In this day and age of open data (open gov), the lat/lon information should be made available. May police departments are starting to provide their crime data (not just collisions) via open data portals. Check out http://data.vancouver.ca/datacatalogue/crime-data.htm as an example.
Dave, Thanks for your comment. I have shared your views with the data holder for their consideration. Karin from the open-ouvert team
Hi Dave, The following response has been posted on behalf of the data owner: "Thank you for your comment. The National Collision Database is in the process of trying to collect and report on additional data elements including ‘Collision Location – Geocode’. However, at this time the jurisdictions responsible for collecting these data have not been able to collect the location data on a sufficiently complete basis, so we are unable to compile and release location data at this time." -Tasha, the Open Government Team
will there be a way to search for vehicles themselves? ie search for the VIN to find out if a car was in an accident?
Hello, Thank you for your question. I have sent this to the data owner for clarification. Thanks, Tasha, the Open Government Team
When will you be adding 2014,2015 and 2016 data. This is an excellent data base. I would love to see GPS coordinates added.
James, This from the data holder: Transport Canada is in the process of updating the dataset to include up to and including 2014 data. More current data is not yet available in a consolidated and complete state. Please check back for updates at a later time. Karin - the open-ouvert team
It looks like motorcycle drivers have improperly encoded field P_SAFE in many cases: I have tried to see statics on accidents involving motorcycles, to see how type of protective gear correlates with the severity of injury. Unfortunately, the information in the database seem to be incomplete. I made a subset of data, selecting the rows where (P_PSN=="11"P_PSN=="12"P_PSN=="96") & P_ISEV!="N" & P_ISEV!="U" & V_TYPE=="14" & P_SAFE!='UU' & P_SAFE!='QQ' & P_SAFE!='NN' & P_SEX!='N' & P_SEX!='U' And the summary of the resulting distribution of P_SAFE record shows following (replacing codes with values from the data dictionary): No safety device used or No child restraint used 2063 Safety device used or child restraint used 2282 Helmet worn For motorcyclists, bicyclists, snowmobilers, all-terrain vehicle riders 57773 Reflective clothing worn For motorcyclists, bicyclists, snowmobilers, all-terrain vehicle riders and pedestrians 37 Both helmet and reflective clothing used For motorcyclists, bicyclists, snowmobilers, all-terrain vehicle riders and pedestrians 0 Other safety device used 86 No safety device equipped e.g. buses 4468 Data element is not applicable e.g. dummy person record created for parked cars 0 Choice is other than the preceding values 36 Unknown e.g. applies to runaway cars 0 Jurisdiction does not provide this data element 0
The collision level data includes the "year", the "month" and the "day of week", but there is no "day of month" why is that?
Indeed, I fail to see why we can't get the exact day of each accident, this would be VERY useful.
Hi Yves, we have contacted the data owner regarding this and will provide an update as soon as we can. Stay tuned! Momin, the Open Government team.