Category: Data

Talking about Divvy data with WBEZ’s Afternoon Shift

July 31, 2013 / / 0 Comments

Audio from 3-4 PM segment. Skip to ~30 minutes in to hear the data portion of the show. 

WBEZ Afternoon Shift host Charlie Meyerson had me on the show Monday to talk with him, Gabe Klein, and some fellows at Data Science for Social Good (DSSG, a University of Chicago-based program) about the trends that Divvy bike sharing data is showing. Here are trends I mentioned or was prepared to mention:

* 31 stations installed since Friday, July 19 (10 days since Monday, July 29), 28 stations installed since I wrote my post on Streetsblog Chicago about 67 memberships per day, so I predict that the daily rate of new annual members has increased.

* Membership enrollment is still concerning to me: From July 22 (when I wrote my post about enrollment rate) to yesterday (Sunday, July 28), membership enrollment rate dropped to 61 memberships per day, even as all these new neighborhood stations emerged. This brings the post-launch average to 66 from 67.

* If you look at Top 10 starting and ending stations, there’s about 90% crossover, meaning the Top 10 starting stations for trips are basically the same as the Top 10 ending stations. There’re slight changes on the weekend, with Lincoln Park (the park) and the Lincoln Park Zoo getting into the Top 10. (See table below.)

* During the week, Union and Ogilvie Metra stations get into the Top 10, and disappear from the Top 10 on the weekend. This may suggest that commuters, not tourists, are making these trips. But people make tourism trips from the suburbs on weekdays as well.

* Trips by member type: 71% are taken by 24 hour pass holders. This is down from 75 to 73, so this means that the share of trips taken by annual member holders is increasing. This is because of two things happening: some people who bought a 24-hour pass to test the system have converted to being an annual member, and others who waited for a station to come to their neighborhood have bought a membership. I’m just hoping that membership enrollment picks up to reach the high rate of installations.

I’m personally interested in the route data. I’m interested in who’s biking where and when. This is information we’ve not collected well in the past. My Streetsblog Chicago partner John Greenfield wrote about other data trends Scott Kubly discussed at last week’s Complete Streets Symposium.

Dock surfing during a Divvy social ride last Thursday. Photo by Jane Healy.

[table id=9 /]

Download this data

Divvy data from May 29 to July 28 (.xls): includes member enrollment, number of trips taken by annual and 24-hour pass holders, and top 10 starting and ending stations.

If you’re looking to contribute your expertise to the “Divvy data project” (okay, such a thing doesn’t really exist), then check out the Divvy Data Document I started after discussing Divvy data at a July OpenGov Hack Night.

Revealing driver behavior on Clark Street with a radar gun

July 30, 2013 / / 2 Comments

People prefer to cross Clark Street at Menomenee Street in groups of unacquainted individuals.

This is a more detailed post of the one at Streetsblog Chicago.

On the overcast morning of Friday, May 4, 2012, I recorded the speeds of 412 cars at four locations along Clark Street in Old Town and Lincoln Park for 15 minutes at each location. I missed counting the speeds of 42 cars. The embedded map shows the locations and some basic statistics.

What did I find? There’s a relationship between street width and the speed people drive. The highest speeds were found on the widest portions, and the lowest speeds on the narrowest portions. However, this basic study is far from scientific. A better study would record the locations simultaneously (necessitating 4 radar guns), calibrated equipment, consistent training for the researchers on data collection methods, a longer recording duration, and comparison to a control street that had a uniform width at four locations.


View Radar gun places on Clark Street in a larger map

1. Southbound Clark Street at Germania Place

My assistant and I set up the radar gun and camera immediately south of Sandburg Terrace and pointed the radar gun at people driving southbound on Clark Street between a row of parked cars at the concrete median (pedestrian refuge island). Classes would start soon at the Latin School on the east side of Clark Street. Compliance with state law requiring drivers to stop for pedestrians in the crosswalk was weak, to say the least, but compliance wasn’t explicitly measured.

  • Average speed: 17.21 miles per hour (MPH)
  • Maximum speed: 30 MPH
  • Cars measured: 151
  • Speed limit: 30 MPH
  • Drivers exceeding the speed limit: 0
  • Width: 224 inches (from west curb to pedestrian refuge island)
  • Effective width: 140 inches (excludes parking by subtracting 7 feet)
  • Crashes: 35, of which 4 were bicycle, and 3 were pedestrian.

Only one car-car crash (actually a 3 car crash) produced an injury. What’s interesting about this location is that in a lot of the crashes, the cars were traveling in the same direction. There’s a lot of school drop off and pick up activity here for Latin School of Chicago students, so it could be that many people are pulling away from the curb to merge into traffic and collide.

2. Northbound Clark Street at Menomenee Street

  • Average speed: 30.83 miles per hour (MPH)
  • Maximum speed: 50 MPH
  • Cars measured: 121
  • Speed limit: 30 MPH
  • Drivers exceeding the speed limit: 53.72%
  • Width: 395 inches (from east curb to dividing line). This includes the parking lane but no cars were parked within 50 feet, north and south, of the measurement location.
  • Crashes: 20, of which 2 were bicycle, and 1 were pedestrian. Many of the non-bike and non-ped crashes involved a parked car or taxi. The only injuries experienced were by the 2 cyclists and 1 pedestrian.

3. Northbound Clark Street at Lincoln Park West

We stood on the “pie” (traffic island) that separates northbound Clark Street traffic from northbound Lincoln Park West traffic to measure the traffic driving on Clark Street between the pie and the concrete median separating it from southbound Clark Street.

  • Average speed: 25.60 miles per hour (MPH)
  • Maximum speed: 40 MPH
  • Cars measured: 58
  • Speed limit: 30 MPH
  • Drivers exceeding the speed limit: 27.59%
  • Width: 252 inches (from concrete median curb to west curb on the pie)
  • Crashes: 4, of which 1 was bicycle, and 2 were pedestrian.

4. Northbound Clark Street between Lincoln Park West and Dickens Avenue

This location is 125 feet north of the previous location.

  • Average speed: 22.54 miles per hour (MPH)
  • Maximum speed: 35 MPH
  • Cars measured: 58
  • Speed limit: 30 MPH
  • Drivers exceeding the speed limit: 2.44%
  • Width: 264 inches (from east curb to dividing line).
  • Effective width: 180 inches (excludes parking by subtracting 7 feet)
  • Crashes: 0

Me measuring speeding drivers on Clark Street with the speed gun, my clipboard and paper, and a GoPro camera to record the speeding drivers and the results on the speed gun. 

Bike Walk Lincoln Park’s proposal

In 2011, Michelle Stenzel and Michael of Bike Walk Lincoln Park published a document to “Make Clark a Liveable Street“. The first two pages show an aerial photo of the same section of Clark Street where I measured automobile speeds, North Avenue and Armitage Avenue. On the first page, existing conditions are laid out. The second graphic shows proposed improvements.

At Menomonee Street, measurement location 2, the document says “pedestrians must cross 6 lanes with no safe haven”, a width of just under 66 feet. In the later pages, the first existing condition is blatant: “Wide lanes of auto traffic moving at speeds in excess of the speed limit”. My analysis in May demonstrates this.

How does BikeWalk Lincoln Park propose to “transform this stretch from a car-oriented ‘super-highway’ to a people-oriented liveable street”? By installing protected bike lanes, putting the street on a diet, and installing new and well-marked crosswalks among other ideas.

Width and speed summary

Ordered by location:

  1. 224/140 inches. 0% of drivers exceeded 30 MPH speed limit
  2. 395/395 inches. 53.72% of drivers exceeded 30 MPH speed limit
  3. 252/252 inches. 27.59% of drivers exceeded 30 MPH speed limit
  4. 264/180 inches. 2.44% of drivers exceeded 30 MPH speed limit

Ordered from narrowest to widest to see how width relates to speed:

  • 224/140 inches. 0% of drivers exceeded 30 MPH speed limit
  • 264/180 inches. 2.44% of drivers exceeded 30 MPH speed limit
  • 252/252 inches. 27.59% of drivers exceeded 30 MPH speed limit
  • 395/395 inches. 53.72% of drivers exceeded 30 MPH speed limit

Notes

Crash data is within 100 feet to avoid the overlap of the final two locations, which were 125 feet apart. Crash data comes from the Illinois Department of Transportation for 2005-2010. The Bushnell Velocity Speed Gun was borrowed for this analysis. The radar gun was filmed to show a speeding car and its speed simultaneously. The video below shows a driver traveling at 50 MPH in a Children’s Safety Zone (as it’s within 1/8 mile of a park, Lincoln Park, making it eligible for automated speed enforcement).

Curiously, no traffic counts have been collected on Clark Street near any of the count locations.

View the video on Vimeo.

Screenshot of traffic count website. Go to the Traffic Count Database System and search for “1700 N Clark Street, Chicago, IL” in the map. 

Divvy memberships growing at very slow rate

July 24, 2013 / / 36 Comments

Chart showing the progression of annual member sign-ups. 

The day after Divvy – a bike sharing system in Chicago operated by Alta Bicycle Share – started signing up members, enrollment dropped by 80.2%. The next day it dropped by 57.9% and then 55.7% after that. The progression was 732 in the first day, 145 in the second, 61 in the third, and 27 in the fourth day. Since day two, daily enrollment has never exceeded 121 sign-ups in a day.

The Bike to Work Day Rally on Friday, June 14, had almost no impact: there were 6 sign-ups that day, with 6 sign-ups the day before. This was the first time that a station was visible to the public and Divvy staff were out there talking to people and allowing some test rides. The next day, however, there were 12 sign-ups. Even launch day, June 28, was weak, especially given that the new system was given a lot of attention that day and weekend in the press. The Monday after Friday’s launch saw more than launch day.

In a system that has so far focused on a few stations in neighborhoods (like West Town/Wicker Park, South Loop and Lincoln Park), this might not be surprising. Nor is it surprising that memberships were low from the period enrollment opened to the first station being installed – because there was nothing out on the street to catch people’s attention and you had to know about it by being told, online or from a friend.

“What is this?”

I expected, then, that memberships would jump once the system went live, to at least a rate higher than the period when membership was open but there were 0 stations installed. But that hasn’t happened. If the rate of new annual members doesn’t start increasing as stations start increasing, I will be very concerned. Currently, most trips are taken by 24-hour pass holders, and the most popular stations are near the lakefront, telling me that the system is used mostly by (confused) tourists.

Riding a Divvy bike on Dearborn Street.

I could, of course, try to compare us to New York City’s rapid explosion in annual member sign-ups for Citibike, also run by Alta Bicycle Share. I’m sure readers would poke holes in that comparison, rendering any argumentation here useless. Now that there are 75 stations are in place, the rate of post-launch annual membership enrollment should be vastly higher than the rate of pre-launch annual membership enrollment. The period when there were 0 stations had a higher rate of enrollment than the period that followed it during which stations were being installed.

Recap

Pre-launch, days 1-16 of enrollment (with 0 stations)
1,152 memberships, 36.5%, average of 72 members per day

Pre-launch, days 17-30 of enrollment (with 1-68 stations)
320 memberships, 10.1%, average of 22.9 members per day

Subtotal: Pre-launch, 30 days of enrollment (with 0-68 stations)
1,472 memberships, 44.7%, average of 49.1 members per day

Post-launch, days 31-55 of enrollment (with 68-75 stations)
1,685 memberships, 53.4%, average of 67.4 members per day

Total: days 1-55 of enrollment (with 0-75 stations)
3,157 memberships, 100.0%, average of 57.4 members per day

Appendix

View the membership data for yourself (XLS).

P.I was told two weeks ago that marketing for Divvy would soon begin on Chicago Transit Authority bus shelters and “City Information” signs, both advertising infrastructure operated by JCDecaux under its contract with the City of Chicago. I think this will have minimal impact, but it’s definitely worth putting out there.

Tell me I’m wrong with my Parking Meter Deal Part Deux calculations

April 30, 2013 / / 16 Comments

A parking meter in Milwaukee, Wisconsin, displays the word “fail”. Photo by Jeramey Jannene.

I want nothing more than to believe Mayor Rahm Emanuel has created a good deal but I believe his own parking meter deal is just as ridiculous as the deal – from Richard M. Daley and 45 aldermen – that preceded it.

Rahm’s deal changes none of what Chicagoans abhor about the current deal, which include:

  • It scheduled many price increases, without offering the buyer, those who pay to park, any additional value. Value could come in the form of a parking benefit district, where the revenues pay for local infrastructure improvements.
  • The city gets none of the revenue (it collects fines, though).
  • It costs us more than we ever expected (disabled parking placard, reimbursement for street closures, road work, and festivals).
  • It removes control from the city administration and aldermen over our streets. Thanks goes to Active Transportation Alliance for pointing this out in their excellent June 2009 original report (since retracted and revised) in which the organization said, “As a result [of the lease], planners and neighborhoods have lost control over one of their most powerful urban planning and revenue generating tools.”

It changes nothing that policy makers dislike about it:

  • We can’t implement dynamic or market or congestion pricing, unless the revenues for CPM stayed the same or were increased (although this would have to be negotiated).
  • It throws another cog into the city’s plans to expand bike lane mileage. We’re already having a difficult time with merchants not wanting to lose parking in front of their store, despite all the evidence pointing to bike lanes increasing revenues. To make way for a bike lane, the metered parking space has to be moved to an equally valuable spot within the same Parking Region. The alderman has to get involved and it’s not an easy process.

Rahm’s deal, which the city council must approve as an ordinance, doesn’t help Rahm’s priorities.

The Active Transportation Alliance report said, “This lease agreement [from 2009] compromised the city’s ability to adjust parking policy; because of the agreement terms, meters will be the primary consideration in the planning of our city streets. Everything else, from traffic flow to pedestrian, bicycle and transit facilities may only be considered after meters and their corresponding income has been considered.”

Rahm’s new deal doesn’t change that, but in fact will likely give CPM the same or more revenues under the plan. It will reduce the chargeable hours by 12 hours on one day (the newly free Sunday) and increase by 1 hour at more valuable times (weekday and weekend evenings) in areas that charge $2 and $4 per hour, and is increased by 3 hours at the same times in areas that charge $6.50 per hour. I’ve attempted to estimate how much more revenue with the spreadsheet below.

The city isn’t saving $1 billion – it hasn’t spent that money and there was no surety that it would; the press release acknowledges this, calling them ” estimated future charges”. The point here is that CPM and the city have agreed on how things like street closures and disabled parking placards will be paid for (by the city). CPM isn’t going to agree to any deal that reduces the value of the company to its shareholders.

No one asked to have free parking on Sunday. No one asked to have free parking on any day. Sunday is the day when people drive the least! If anyone deserves a break, it shouldn’t go to a small segment of the popular (“Sunday churchgoers”, Rahm said, acting as if they’re being harmed, and excluding churchgoers who don’t attend on Sundays), but to everyone who had to pay more than the parking space was worth and anyone who couldn’t get a bike lane in while people are being doored left and right.

Why else is free parking a bad idea? The experts at Active Transportation Alliance wrote:

Underpriced curb parking is a hidden source of traffic congestion and stimulates the most inefficient form of urban transportation. Underpriced parking encourages drivers to cruise for cheap parking, which harms everyone’s health and safety, slows down automobiles and buses behind the cruiser, and provides little benefit to the cruiser. It is a danger to bicyclists and pedestrians because cruisers focus on finding the right spot, not on whether a pedestrian is crossing the street.

It’s this last point, the lack of focus on anything but the parking spot, that is believed to be the cause of a cyclist being severely injured last week on Milwaukee Avenue.

Just like Daley, Rahm didn’t consult the one alderman whose ward might be affected most (it’s unknown if any aldermen were consulted). If this trend of the current city council being the most “rubber stamping” in all time (by my favorite local blogger Whet Moser), I predict it’ll be passed.

Calculations

[table id=8 /]

Since the number of spaces doesn’t change between the old and new scenarios, there is no need to calculate the total $ per space per region. Revenue estimate assumes the space is always occupied. In the new scenario, proposed by Rahm Emanuel and CPM, all spaces not in neighborhoods have become slightly more valuable, enough to more than make up for the reduced value of spaces in neighborhoods.

Updated May 3, 2013, 15:51 to add a link to the current version of Active Transportation Alliance’s parking meter report and to say that it replaced the original report. 

Results of my personal #editathon this weekend

April 21, 2013 / / 0 Comments

I added a lot of parking lots to OpenStreetMap this weekend, but I also added the Willow Creek Community Church (South Barrington, Illinois) parking lots, driveways, buildings, and retention ponds. I’ll let my before and after screenshots show you what I did. Ian Dees, local organizer for Chicagoland OSM data – he has other roles, too – said there’s an application that can generate these images automatically.

View in OpenStreetMap now.

I’m also writing a draft tutorial on how to convert GIS data stored as a shapefile to a format you can import into AutoCAD. GRASS will take a .shp and convert it to .dxf (a geo-aware CAD file).