After almost half a year of limiting vehicles on King St, many are still wondering if the pilot has been a success or a failure.
The King Street Transit pilot was a bold move. By shutting down the passage of all private vehicles along the busiest section of the arterial street, city officials took a dramatic approach to moving people through the city, and, as they put it, “support economic prosperity.”
Whatever its intention, the backlash against the pilot was immediate.
Many drivers weren’t immediately aware of the project and drove through King as normal. Other more obstinate drivers ignored the new rules entirely and were promptly ticketed. Whether it was confusion or stubbornness, as of the beginning of April Toronto’s finest have written over $500,000 worth of tickets
along the route. In addition to problems on King, many commuters reported that Queen Street and other nearby roads became badly congested, and conservative critics labeled the entire pilot a kind of “war on the car” that was directly aimed at commuters from the suburbs.
What’s worse is no one seems to know whether or not it’s working. A study from University of Toronto researchers show that the pilot has caused “dramatic improvements in travel times and reliability” while another from a University of Ryerson professor show that the improvements have been “modest at best.” These discrepancies are at least a little odd, given that both studies are using the same data.
Ultimately, the sample size for the data being analyzed is still very small. The pilot has been in effect for less than half a year, and the data provided by the TTC has been used by both its advocates and critics to serve their purposes. The shouts from both sides on social media also seem to suggest that many have already made up their minds about the pilot’s efficacy. This is problematic insofar as the decision to expand, modify, or scrap the pilot altogether should be driven by data, not public sentiment.
A new source of information
In January of last year ridesharing company Uber launched Uber Movement, a data-hosting service where aggregate data from their vast network was made publicly available. The initiative was aimed at city officials, urban planners, and anyone who would be able to use the data to discover patterns in traffic, compare travel times, and ultimately make data-driven decisions to improve the future of urban mobility.
Last week Uber released data for Toronto, and we immediately knew what questions we wanted to ask.
Step One: The Data
To protect their users’ privacy, Uber is aggregating the data according to different source and destination zones. These correlate to Toronto’s neighbourhoods.
Toronto’s Neighbourhoods are geometrically very similar to Uber’s, but they have different names
Because of this level of aggregation, it’s not possible to track a specific route (such as King at Strachan to King at Jarvis) but rather get a sense of how traffic in general is flowing throughout the city. We decided that rather than try to pin down what was happening on King Street specifically, we’d focus on directional traffic and ask a very basic question:
Has the King Street pilot had an impact on west to east travel times?
To dive into the data, we started by downloading and combining the data for all available time periods (Q1 2016 through Q1 2018). This provided us with three distinct data sets:
- Travel time by hour of the day
- Travel time by day of the week
- Travel time by month
These data sets provide us with a historical baseline of data that we can use in our analysis. In order to look up the Uber zone for different addresses, we turned to Toronto’s open data catalogue. By performing a spatial join
on Uber’s Neighbourhood data and Toronto’s One Address Repository
, we can create a linked data set
that ties the two together, which lets us effectively translate Uber’s data into terms that are more familiar.
To track west to east traffic, we chose 13 neighbourhoods in Toronto and mapped them to Moss Park, the point at which Carlton, Gerrard, Dundas, Shuter, Queen, Richmond, Adelaide, and King all flow into a single zone.
The 13 Toronto Neighbourhoods and Moss Park
Step Two: The Analysis
After the wrangling, we still had a lot of data to sift through. When visualizing large data sets, we use tools like Tableau to help drive immediate insight. To keep things simple, we started by displaying monthly aggregates of the travel times from January 2016 to March 2018. In general, there is a downward trend in mean travel time.
To plot a more general trend, we also see minor improvements in travel time from December 2016 to December 2017.
And a relative decrease over three years.
The changes in travel time aren’t dramatic, and this is likely why there’s been a degree of uncertainty about the King Street pilot’s success. To critics this data could represent a change that’s not worth the effort; to the advocates it might show a clear downward trend. Either way, it seems clear that the pilot is not actively slowing down travel times across the city, and that’s an important data point.
It may be an unconventional metric of success, but the fact that the pilot didn’t have an immediate negative effect on travel times throughout the city is significant
These findings are based on aggregations. As we dove into the data at more granular levels, we occasionally saw travel times between locations increase during peak hours. Although the general trend has been static or decreasing travel times, these results aren’t universal. The data provided by Uber is not a magic bullet but another layer, another dimension, to the King Street pilot; one that helps us understand the contours of the issue a little better and question it more thoroughly. We encourage everyone from civic leaders to high school students to dig into the data, slice and dice it in different ways, and use it to paint a better picture.
The world of personal data is no longer a lawless town in the Wild, Wild…
Mark Tunbull, managing director of Cambridge Analytica (CA), was well aware of what the public…