Ride-sourcing (TNC service) and transit in Shanghai
1. Friend or Foe?
App-based, on-demand ride services
(ride-sourcing) and transit in Shanghai
UC Berkeley
Ruoying Xu
PhD student at the Department of City and Regional planning
Yiyan Ge
Concurrent Masters student at DCRP and Transportation Engineering
2. Friend or Foe?
App-based, on-demand ride services
(ride-sourcing) and transit in Shanghai
Transportation Planning & Urban Data Science
UC Berkeley
Ruoying Xu
PhD student at the Department of City and Regional planning
Yiyan Ge
Concurrent Masters student at DCRP and Transportation Engineering
3. What question are we trying to
answer and why?
How do we approach the
question?
How do we implement the
approach?
What can we do with the findings?
OUTLINE
5. Why do we care?
Because how we travel
shapes our experiences living
in the cities.
6. We make choices
• Demand
• Travel mode
• Travel time
• Location
preferences
Choices have
consequences
• Traffic
• GHG emission
• Land use patterns
TNC
Equity and Access
7. Is ride-sourcing (TNC) competing with
transit in cities?
Traditional approach:
Whether people actually switched from a transit
mode that they were previously using to the
new mode TNC for the same trip purpose?
11. DATA
Trip data from Jan. to Oct., 2015, provided by Didi Kuaidi
Trip origin and destination
Trip date and time
140,854 samples in total
12. January as the base year: 6098 trips
Total sample size: 140,854 trips
Total TNC trip changes over 10 months
13. Assumption 1
When TNC trip price decreases,
people take more TNC trips,
including trips with transit
alternatives.
14. TNC
TRANSIT INDUCED
DEMAND
If there is a
reasonable transit
alternative available
for the TNC trip OD
[Competition?]
No reasonable
transit alternative
available
Assumption 2
• Low car ownership
(~15%)
• High transit usage
(50% of trips)
• Limited taxi supply
(20 per 10000 ppl)
15. Assumption 2
Origin + destination + day of week +
time of day + transit mode à Google
Map Direction API à transit alternative
Reasonable transit alternative:
• Waiting time < 20 min
• Walking time < 30 min
• Number of transfer at most 1
• Transit travel time / TNC travel time
ratio <= 2
16. Is ride-sourcing (TNC) competing with transit in cities?
Individual level:
Assumptions on travel behaviors
17. Is ride-sourcing (TNC) competing with transit in cities?
Individual level:
Assumptions on travel behaviors
Hypothesis:
e.g. people are more likely to use TNC service for
short-distant trips
18. Is ride-sourcing (TNC) competing with transit in cities?
Individual level:
Assumptions on travel behaviors
Hypothesis:
e.g. people are more likely to use TNC service for
short-distant trips
EXPECTED
differences and
changes in % of
TNC trips that can
be reasonably
replaced by transit
19. Is ride-sourcing (TNC) competing with transit in cities?
Individual level:
Assumptions on travel behaviors
Hypothesis:
e.g. people are more likely to use TNC service for
short-distant trips
EXPECTED
differences and
changes in % of
TNC trips that can
be reasonably
replaced by transit
OBSERVED
differences and
changes in % of
TNC trips that can
be reasonably
replaced by transit
20. Key Questions
In what circumstance,
ride-sourcing service is more competitive
with transit?
When:
1. the trip distance is short?
2. the transit alternative is bus-only?
3. the trip takes place during peak-hour?
27. % of TNC trips with metro-only or bus-only alternatives
that can be reasonably replaced by metro or bus
28. Takeaway
Ride-sourcing is more likely to be
competing with transit:
1. when it is a long trip
2. when the transit alternative is metro
Prices affect different types of trips
differently
There is strong indication of induced
demand
30. Transportation planning policies that are grounded in
neither theories nor evidence
Lagging transportation planning policies that respond to
the past
No RIGHT process
Correlation is fine too
Collaborations between data owners and
planning agencies
Responsible and responsive transportation
planning policies
Transportation Planning
Urban Data Science