2018-2020 ValleyBike-deep dive analysis-Amherst CollegeDSC WAV: ValleyBike
December 11, 2020 Demo
PI:Nicholas J. Horton (Amherst College)
Team:Umaimah Ahmed, Kenny Chen, Graham Chickering, Chris Chukwura, Maria-Cristiana Gîrjău, Konstantin Larin,
Maggie Wu, Jessica Yu
Introduction
Goals
1.Find new optimal location for future bike stations (docks)
2.Ridership patterns in low-income communities
3.Crafting ‘stories’ for future sponsorships
Available Datasets
Data Infrastructure
●500 daily files in total (28 June 2018 -5 October 2020)
●To import a single day of data: valleybikeData::import_day()
○Parameters:day, return, future_cutoff
●To import a month of data: valleybikeData::import_month()
○Parameters:month, return, future_cutoff
●To aggregate the trips: valleybikeData::aggregate_trips()
●To aggregate the users: valleybikeData::aggregate_users()
Data Summary
Where are bikes going?
●System Summaries
●Station to Station Activity
●Trips w/ no stations
●Heatmaps
Station Summaries
Heatmaps
GOALS
1) Visualize where hotspots are in any area
2) Compare where people are frequenting by month
3) Build tools so users can easily look at any area
Three Julys in Amherst
Northampton
Plot Anywhere and Everywhere
Interactive
Leaflet Plots
GOALS
1) Visualize where riders are traveling from station to station
2) Provide additional information on unique routes users are taking
3) Make it easier to create these plots quickly
3) Look into unique user stories
Florence Center to
Florence Center
Jackson Street to Jackson
Street
Florence Center to
Jackson Street
Basketball Hall of
Fame
NA Trips
●Some rides don’t end at a specific station
Potential Reasons:
●Repairs
●Secondary Stops
●Malfunctions
Proportion of Rides That Don’t End at a Station
●2-3% generally
●Some days over 5% and close to
10%
Close Look
●Sample of 16 trips from NA from Springfield
●Dates: June 2019
●Many relatively close to Kenefick Park Station
Interactive Map
Secondary Stops
Goals
1) Identify rides that might be using the secondary lock mechanism in order to make a stop
somewhere that is not a station, and then continue along their ride.
2) Analyze these rides to find any patterns, e.g. most common stopping locations.
3) Plot some specific rides on a map, along with their corresponding stops, in order to visualize where
the user is going and where the secondary stops are taking place.
4) Provide additional information on where the starting and stopping stations are, along with the stop
duration and the total ride duration.
User 1
User 2
User 3
User 4
All Secondary Stops
Northampton Stops
Springfield Stops
Interactive Secondary Stops
Who is riding the bikes?
-Overall summary of demographics
-Individual user activity
User Demographics: Age
User Demographics: Age II
User Demographics: Gender
User Demographics: Age and Gender
●Male users 18-23 and Female users 23-30 have the shortest
median trips
●Older users consistently have a bit longer trips
User Demographics: Location
●Location represents user address when
setting up an account
●Other includes both smaller locations in
the Pioneer Valley and places outside of
it, e.g Boston and New York.
●Other -fewer trips per user
Weekend and Workday Trips
Shorter trips on Workdays
Weekend and Workday II Trips
Most Popular Stations by User Location I
Most Popular Stations by User Location II
Most Popular Stations by User Location III
User Stories
Publicly-Available Utilities
●valleybikeData package to hold the datasets; valleybike package to hold utilities
●Data access utilities:
○get_trajectory_data()
○get_distance()
○get_duration()
●Data exploration utilities:
○get_station_summary()
○identify_trips()
○plot_trip()
○get_stops()
Future Work
●Analysis of short trips (scooter competition?)
●Low-income community exploration
●Forecast bike usage
●Characterizing trips (running errands, going to school, etc.)
Acknowledgments
●NSF grant (The Data Science Wrangle-Analyze-Visualize project: Experiential Learning with Local
Community Organizations), #1923388 (Nicholas Horton Amherst College PI)
●Wayne Feiden, City of Northampton
●Andrea Dustin and Ben Baumer (Smith College)