<< LINK TO LATEST DASHBOARD >>
I’m a life-long environmentalist and the promotion of stewardship of our shared planet is one of the driving forces of my life. I work in the public transportation industry, I live car-free, I bike and take transit everywhere, I maintain a plant-based diet. Realizing I could do even more to promote a sustainable vision of our shared future, I decided to run for advisory neighborhood commissioner (ANC) here in central Shaw (6E01). I am using my seat to promote safe and sustainable transportation, affordable housing, and data-driven decision making.
I’ve been active in the ANC 6E Transportation Advisory Committee (TAC) since moving to Shaw in late 2019, advocating for spot improvements to unsafe intersections and bringing my background in transportation engineering and city planning to the hyper-local level to help neighbors advocate for their needs. One dimension of this advocacy is using data to tell stories. Saying “cars speed down my block” can only resonate with elected officials so much; saying “there have been 100 speed-related crashes in my neighborhood over the past 3 years” paints a much clearer and more actionable picture of a problem ready to be solved.
The DC Crashes Data Viewer Story
Back in December, before getting sworn in as ANC, I found the Crashes in DC dataset on DC’s open data website. Immediately I knew I wanted to download this data and develop a way to get the relevant information into the hands of ANC commissioners and neighborhood traffic safety advocates.
One of the first things I noted was the crash data points were annotated with the ID of ANC where they occurred. I wanted to get even more fine grained, so I pulled the dataset up in qGIS and overlaid it with the ANC Single Member District (SMD) boundaries shape data. A quick geoprocessing command later I had a crashes dataset that included the SMD_ID as well.
Next, I realized that seeing points on a map was not the best way to visualize the data: many points overlapped with one another, obscuring the biggest trouble spots. I scanned the data fields available on the Crashes dataset and found a BlockKey data element that matched the BlockKey field on the Roadway Block GIS data file available from the open data site.
Now I had everything I needed: historical data of crashes on DC roadways, a shape file of DC roadway blocks, and a shape file of DC’s AND SMDs. For me the natural first step is putting data into Tableau, a data viz platform that’s easy to use and free for personal projects, including free online hosting of data visualization dashboards. After some hours exploring different ways to pivot and display the data, the current DC Crashes Viewer was born.
What About DDOT’s Vision Zero Portal?
Since starting work on this data viewer, DDOT released their own crashes data visualization. This great new tool was recently featured on GGWash. While all the data is the same, there are two features of my dashboard that this portal doesn’t do: 1) break out crashes by single-member district; 2) group crash counts to the roadway block level. Really this second one is the most important to identify trouble spots and places to focus activism/advocacy efforts.
What Can This Dashboard Do?
The primary purpose of this dashboard is to help provide data on number of crashes and traffic fatalities by ANC, SMD, or roadway block/segment. This specific data can then be incorporated into ANC resolutions, Traffic Safety Assessment requests, and letters to DDOT and other government/elected officials requesting attention to DC’s roadway troublespots. The newest feature allows a user to identify, quantify, and map all the crashes along an entire street, with the data available block by block. This feature was requested by an ANC commissioner rightfully noted that major arterials are often the biggest trouble spots and also boundaries between ANCs and SMDs. The mapping of crashes to ANCs is done by the geolocation of the crash, so crashes along any given street can fall on one side or the other of a geopolitical boundary, watering down the numbers and obscuring the urgency of trouble spots.
Use Case 1: Crashes by ANC, SMD
To identify the crashes for a particular jurisdiction, simply select the ANC from the dropdown list on the top left of the main screen. The maps and tables refresh filtered to only show the ANC in question. To focus in on one SMD, click that SMD in the map on the upper left, and the other map and tables refresh to show crashes only in that district.
Example. A recent tweet highlighted concerns about recent crashes in ANC 7C. To pull up data about crashes in 7C, I selected 7C from the ANC drop-down list. Everything refreshes to show only this ANC.
Some things are immediately notable: 1) 7C05 is a real trouble-spot, with the most crashes clearly visible in the map on the upper left; 2) The worst individual block is in 7C06, the 934-1099 block of Eastern Ave with 102 crashes in the selected time period, as notable in the table and map of Crashes by Block. The heat map in the lower right also shows something notable: crashes appear to have spiked a bit between July and November of 2020, right in the middle of the pandemic.
Clicking on any SMD in the upper-left map refocuses the entire screen on that SMD only. The recent tweet was about crashes on Sherriff Rd NE, which is in 7C04.
This refreshed viz shows the 5100 block of Sherriff Rd NE being the biggest problem spot in the SMD. And it’s clear from the heat map that crashes clearly increased during 2020!
Use Case 2: Crashes by Street
The recent death of a cyclist on Michigan Ave NE has brought renewed attention to the dangerous conditions for cyclists and drivers along this stretch of roadway. Using a new feature of the DC Crashes Data Viewer, you can create custom maps and tables of crashes along a given roadway. I shared how to use the Crashes by Street feature in a series of tweets, resulting in illustrations such as this;
Combined, drivers along Michigan Ave have crashed over 700 times since 2015, including 99 crashes on the 100-379 block NE.
Featured on WUSA9!
Since taking this data viz to Twitter, I was contacted by a reporter from WUSA9 who gave me the opportunity to speak on camera about the work i’m doing with DC crash data. Check it out! Thanks, Matthew and WUSA9!
I plan to continue making enhancements to my crashes dashboard while also sharing the data with additional local elected officials and traffic safety advocates. I’d like to automate the crashes data it refreshes automatically, which might require moving it from Tableau to maybe R Shiny, which I have yet to learn. I’m also excited to dig into additional DC datasets, such as the new data dump of 311 requests by year, available on the open data website. I’ve started exploring the 311 data for 2020 already.
I’m also looking to create a regular Urban Data meet-up, bringing together data science/viz folks and local elected officials and advocates so we can share ideas, datasets, techniques, and moral support while building tools that transform raw data into actionable intelligence to help advocate for safer streets and more livable neighborhoods in DC. If you’re interested, please send me an email or follow me on Twitter to look for announcements.