Hudson Bike Share by the numbers

Hoboken, NJ, has been striving to make itself more bike friendly. To that end, it deployed a bike share system in 2015, operated by Hudson Bike Share. This system fills a useful niche, providing a convenient option for the last mile of transportation — quite literally in Hoboken’s one square mile.

Hudson Bike Share (what I like to call “HUBS”) makes their trip data available on a quarterly basis.

(Note: The dataset accidentally omitted the end of each quarter. These figures will be updated when the complete dataset is made available.)

Background

A Hudson Bike Share bicycle docked at the Pier A Station.

HUBS uses NextBike technology. Like the Social Bicycles used in Hoboken’s 2013 bike share pilot program, this technology is different from systems like Motivate (used next door and across the river by CitiBike). Each bike has its own GPS and locking system. This means that stations are only geofenced areas. HUBS has installed racks specific to its bike, but they are not required to finish a trip the way CitiBike requires the bike be docked in a rack. The bike only has to be within the geofenced location and can be locked to itself (a bar goes through the fork and front wheel). This is especially helpful for riders because there is never a situation where a dock is full and they can always end the trip exactly where they want to.

This also means that a trip can be ended anywhere. The dataset provided includes these trips, with the start or stop station being listed as the bike itself. Unfortunately, with the provided data, there’s no way to know where such trips started from or stopped.

Overview

From October 2015 through December 2016, riders took 169,482 total trips, with a clear peak during warmer weather.

The total number of bikeshare trips per week, from October 2015 through December 2016.

The total number of bikeshare trips per calendar day. Each day is shaded by its percentage of the maximum daily usage. (The provided dataset omitted the last day of each quarter.)

Weather

A series of metal poles with holes at the top lined up along the waterfront walkway sticking up through snow. The walkway is partially cleared of snow. Hoboken Terminal is visible in the background through falling snow and sleet.
An empty Hudson Bike Share station during a nor’easter on March 14, 2017. The system was shut down for the duration of the storm and the bikes placed in storage.

As expected, the weather has an impact in ridership, but even on a 16°F, snowy day there were intrepid riders out-and-about. In fact, the only days that had zero trips were days the system was not operating because of storm conditions.

Daily high temperature versus the number of trips that day. The points are shaded by the day’s dominant weather condition: clear, rain, or snow. Weather data Powered by Dark Sky.

Patterns

Looking at the trips by hour of week, there are clear commuting and recreation patterns. Weekdays show rush hour spikes in the morning and evening, while weekends have higher midday volume.

Cumulative trips by hour of week, Sunday, 12am Eastern through Saturday, 11pm Eastern.

Trip destinations

There are currently 34 stations, including the “virtual” ones. The behavior patterns are even more clear looking at the specific destinations by time of the week, and highlight how multi-modal Hoboken is. There is a surge of trips ending at the Hoboken Terminal on weekday mornings, and a surge of trips starting from there in the evenings. The bike share system is a useful companion method to taking the PATH into New York City.

Trip starts and ends at each in-Hoboken station during the week in three-hour snapshots. The green semicircles are trip starts, the red are trip ends. The several stations at Hoboken Terminal have the most trip ends in the weekday morning, and the most trip starts in the weekday evenings. The 14th Street Ferry and Stevens South stations are also prominent. However, the entire system sees trip ends at all times of the week.

Grouping the trip destinations into five categories, Hoboken Terminal, ferry terminal, waterfront, Washington St., and others, we can see how the system is used on a relative basis.

Percent share of each trip destination category per hour of the week. Trips to Washington Street have a fairly even share throughout the week, while Hoboken Terminal shows the morning commute spike. Non-transit stations on the waterfront are relatively more popular on the weekends.

Grouping the trips by “route”, essentially all possible pairs of stations in the system, we can see in more detail where people are coming from and going to. 547 routes have been used, out of 561 possible in 34 stations. This doesn’t include any trips that ended outside of a station area. This also doesn’t tell us where the bikes actually go in between these start and end points, though Hudson Bike Share has released heatmaps using their GPS data in the past.

The waterfront is particularly popular, from 14th Street to Pier A and back, reinforcing the value of the new protected segment of the walkway along the Union Dry Dock facility.

Each row of the matrix shows a route start, and each column is a route end. The ends are grouped by station category. The cells are colored by end station category, and shaded by number of trips relative to the maximum per route. Yellow cells are round trips. The cells with orange borders are routes that have never been used.

The chord diagram shows all of the stations connected by how popular that route is. Hovering over each station isolates the routes that end at it, showing where riders come from to get to a particular station.

Most trips are fairly short, under 15 minutes and typically within the 6–10 minute range. Hoboken’s scale and terrain is perfect for these kinds of quick trips, either as a primary mode for getting around within Hoboken, or as a complement to the PATH or ferries.

Number of trips by duration, grouped into five minute increments.

Trip type

Trips can also be categorized by type, round trip or one-way, which gives some idea of recreation usage versus transportation. They can also be matched with other trips by start, end, and time, showing “joint” trips that presumably are people riding together. So far there have been 5,947 rides with two or more bikes. Without seeing the actual GPS tracks, it’s impossible to say for sure these trips are riders traveling together, but it’s a reasonable assumption. Weekends show a larger share of joint trips than weekdays.

Cumulative trip share per hour of week, categorized by one-way, round, joint one-way, and joint round.

The bikes

As for the actual bikes, they have seen quite a bit of usage. Understandably, some can be soundcite:a little bit creaky. Most bikes have taken at least 500 trips since the system began, with some seeing over 1,000 trips. The system has had 247 bikes in operation so far.


While I would like to get my hands on the actual GPS tracks from each trip and really dive into the geospatial aspects, and have a way to associate trips by a user, there are obvious privacy implications of making that available. Hopefully Hudson Bike Share releases a heatmap showing popular routes.


Trip data from Hudson Bike Share. Weather data from Dark Sky. Charts generated using D3.js. Inline audio powered by Soundcite.