What Portland's Shared Mobility Data Tells (And Doesn't Tell) Us About Micromobility Usage in 2019-2021

On December 16th, the City of Portland filed a press release that I wish more cities would publish - a quantitative recap of their shared mobility program!

Portland’s Public Data

Open Data Portal

The crown jewel of the data released is the city’s “Open Data Portal”, which contains an interactive map heatmap of trip activity, the ability to export summary trip metrics, and information about the most commonly utilized street segments. It’s very fun to play around with, and kudos to the City for publishing this data and making it easily exportable.

PBOT Ride Report: Open Data Portland

Constructive Criticism

While I must commend the City of Portland for publishing their data externally, I would like to note that there are a number of ways in which they could make their data even more “open”:

  • The data does not provide metrics regarding available vehicle counts, which makes it much harder to:
    • understand the evolution of scooter and bike fleets over time, and
    • contextualize ridership numbers (in terms of trips per available vehicle)
  • None of the available data is sliced by Operator, so the public is unable to understand the fleet size/utilization/usage patterns of each provider, only all scooters as a whole
  • The data is sliced by quarter - providing the data at the day, or even hour level would allow for a much more granular understanding of trends:
    • ie. differences in trips volumes and locations by day of week or hour of the day
  • The data provided only deals with trips taken - for full transparency, it would be ideal to also include information about the locations of the scooters and bikes themselves throughout the day, agnostic as to the rides happening on those vehicles

Going Deeper with the Data

After playing around with the Open Data Portal, I knew I wanted to take a deeper dive into the information available, focusing particularly on two things:

  1. Understanding differences in usage between scooters and e-bikes
  2. “Trending” as many metrics over time as possible, to track their evolution over the life of Portland’s shared mobility program and also through COVID’s onset

Take a peek at the visuals I’ve built below to help try and answer those questions!

Summary Metrics

Summary of Scooter and E-Bike Usage in Portland (2019-2021)

Timeseries of Micromobility Trips in Portland

Portland’s Mode Share over Time: Scooters and E-Bikes

Portland’s Distance and Duration Trends

Estimating Vehicle Utilization

Background

Earlier in this post, I made note of the following:

The data does not provide metrics regarding available vehicle counts, which makes it much harder to:	
- understand the evolution of scooter and bike fleets over time, and 
- contextualize ridership numbers (in terms of trips per available vehicle)

I bring this up because not having precise detail on available vehicle counts makes it very difficult to understand how the share mobility pilot is performing at a deeper level. For all the talk about vehicle counts, trip totals, and ride minutes, those numbers are surface level metrics, almost meaningless if they are not being generated in an efficient enough way - context in terms of vehicle supply is crucial.

Said differently - sure, we could get 10,000 rides in a day by dumping 50,000 scooters across the city. But getting 10,000 rides from 4,000 scooters? That’s a much more efficient system, and is indicative of much better product fit, supply/demand balance, and potential economics for operators.

For operators, who need to focus on maximizing their number of trips per available vehicle in order to make money, this number is kind of the holy grail - so I think it’s worth diving a bit deeper so see if we can approximate what this looks like across Portland’s scooter and bikeshare fleets.

Note that due to lack of operator-specific data, we won’t be able to slice this and compare Trips per Vehicle across providers (the most interesting cut of this data) - but we can try to compare the bike and scooter fleets, and see how they have changed over time.

How We’ll Approximate

Bikes

The City of Portland press release itself hints at a fleet of 1,500 pedal-assist e-bikes. The way the sentence is worded leaves it a bit unclear as to whether the 1,500 bikes constitute a completed and “refreshed” fleet, or whether or not they are an incremental addition of bikes into the existing network…

In September 2020, the Portland Bureau of Transportation (PBOT) and Lyft expanded BIKETOWN’s service area to 32 square miles. This included its first expansion into East Portland, and encompassed the Jade District and portions of Lents, Powellhurst-Gilbert and Gateway. A new fleet of 1,500 pedal-assist electric bikes complemented the growing service area.

While I’m still not sure exactly what the correct figures might be (more reason to include vehicle counts in the data being published), I think it’s probably fair to assume that the figure cited in the release, 1,500 bikes, is specified for a reason - so we’ll make the assumption that there are currently roughly 1,500 e-bikes available for ridership around the City.

Scooters

For scooters, we have even less-official figures to work from. As far as I can tell, the City has not published the offical “caps” that each operator’s fleets are limited to, nor the total count of scooters permitted to operated. As such, we have to turn to alternative data sources to ballpark the total size of the scooter fleet.

The best resource I could find is a grant from the Portland Bureau of Transportation and Portland State University to two researchers seeking to “quantify the impacts of scooter operations on VMT”, and “better understand the internal operations and business models of scooter companies”. In the grant summary, they note that:

In the Spring 2019, the City of Portland will be launching a yearlong shared e-scooter pilot program. The city is expecting to have multiple companies providing service in the city starting with 2,500 e-scooters in service.

So - in the absence of official fleet size metrics from the city, we’ll assume there are roughly 2,500 scooters operating in the City.

Visualizing Utilization

With those fleet sizes in mind, we can take the Average Trips per Day metric published by the City and divide it by the appropriate fleet size to compute an estimated Trips per Vehicle Day metric - which we are able to cut by quarter and modality. Take a look at the trends below!

Portland’s Estimated Trips per Vehicle Day