On the release of NYPD’s Crash Data and NYC BigApps’ Vision Zero Challenge

“This is amazing, and more than what we were expecting – the geocoding and integration with the data portal are unexpected and *extremely* pleasant surprises.”

“This dataset allows us to stop second guessing NYPD’s monthly reports, and focus on improving this dataset’s capability. Now, we must push for better collision reporting at the scene.”

Since 2011, members of NYC’s civic hacking community have asked the NYPD to release accurate and machine readable crash data. Today, we commend the NYPD for placing detailed and machine readable crash data online. Additionally, we are ecstatic that Mayor de Blasio shares our vision. We are excited to see the Mayor challenge NYC’s BigApps participants to build tools to achieve Vision Zero.

Why is this important?

In this dataset, we have machine readability, date and time, a latitude and longitude, and collisions that occurred “off street” – for example in a parking lot. Additionally, this data goes back to July 2012. These are all great first steps.

By placing NYC’s crash data online, all community members – civic hackers included – can focus on building platforms and tools to make safer streets. Knowing when and where crashes happen allows the community to look at its actions and see how the streetscape can be improved.

As crashes are preventable, we see this dataset as a fundamental underpinning to make safe streets for all.

How can this data be improved?

The column names seem slightly confused – there are “Number of Persons killed/injured” columns, but no “Number of Passengers killed/injured”. The other six killed/injured columns (“Pedestrians”, “Cyclists”, and “Motorists”) are as expected. Without doing a drill down of the data, it’s hard to tell if this is a transposition of “Passenger” data to “Person”, or if for some reason we lost the “Number of Passengers killed/injured” columns and gained a redundant summary column.
This dataset is missing the total number of vehicles involved and which contributing factor applies to which vehicle.
July 2012 seems a kind of arbitrary cut-off point, considering the old releases were archived back to August 2011.
It is awesome that the NYPD is geocoding these records, but the root of the data is still the intersection. This causes the loss of incident clarity. We continue to ask the NYPD to start providing more detailed location data.
Public safety feeds are only good when they are updated frequently. How often will this be updated? This has yet to be determined.