Hackathon - Lx ped+transit
Hands on!
Why
CML announced that 80% of Lisbon population has access to ZER (Downtown Lisbon) by public transportation.
We are sceptic about this affirmation & map:
- Why using euclidean distances to the transit interfaces? (Poelman and Dijkstra (2015))
- Why distance instead of time?
- How was the population estimated? Did they produce a Dasymetric map?
- Is the access the same at a Tuesday morning and a Sunday afternoon?
What
Let’s share some knowledge, with a practical problem.
Let’s get together, think about this problem (and sub-problems!), and try to have and write a simple solution. The solution might be helpful to other pedestrian + transit problems.
Goal 1
To have a script or procedure we can run and share, open sourced, for a given location/area, time and date, transit operator(s)(?), return the population with access to that location.
It can be a number or a map visualization.
Goal 2
Make it an online map.
Goal 3
Compare Ped + transit with Bike + transit. walk vs. bike
Goal 4 - bonus!
Produce a paper identifying the advances and the caveats of the procedure.
Thoughts:
- Digital pedestrian network (Cambra, Gonçalves, and Moura (2019)).
- GTFS sources for Lisbon (Bok and Kwon (2016)).
- Isochrones polygons - how are they build.
- Pedestrian routes and slopes.
- Pedestrian + PTransit routes (Morgan et al. (2019))
- Pedestrians + equity in access to transit (Pritchard et al. (2019))
- Other literature: Foda and Osman (2010), Malekzadeh and Chung (2019), David S. Vale (2015), David S. Vale, Saraiva, and Pereira (2015), D. Vale (2020)
How
Data we have
- GTFS for Lisbon metropolitan area: timetables, georeferenced routes and stops
- BGRI census 2011 population for AML
- OSM pedestrian network for Carris operation area
(access data here)
Software we can use
Any software, such as programming, GIS, web services, OSM, …
Open source software is preferred.
What to bring
- A laptop?
- Ideas
- A document with your thoughts to share with the others
- A colleague
- Snacks to share
When
Friday, 4th March 2020, 15h30-18h.
Doodle
Where
IST, V0.14.
Thouhgts
Look at tidytransit
From Mateus
Sobre impactos da microacessibilidade das estações de metro/comboio no tempo de viagem: The design trick that could cut 12 minutes off your train commute. David Levinson and Bahman Lahoorpoor, 2019.
Indicadores de acessibilidade (transporte público, a pé e bicicleta) para cidades brasileiras. Disponibiliza códigos em R. Projeto Oportunidades, IPEA. 2020.
Estudo com comparação de tempos de viagem entre modos privados e coletivos. Estimativas considerando diferentes horários e dias de semana/fim-de-semana. Liao et al. (2020)
API do Google Transit para “GTFS Realtime Reference,” em que consta o “OccupancyStatus: The degree of passenger occupancy for the vehicle,” dentre outros atributos que podem ajudar a estimar uma isócrona para o TP mais verossímil - isto será a versão 2.0 do GTFS.
OpenTripPlanner (OTP)| https://github.com/opentripplanner/OpenTripPlanner Open source multi-modal trip planner
From Gabriel
- Tratamento de GTFS para Lisboa: embarques nas rotas e paragens.
r5r
Meanwhile, r5r package was developed by Institute for Applied Economic Research (Ipea), Brazil, and launched in 2021 (Pereira et al. 2021). It compiles locally information of OSM and GTFS for a given area.
It is ver easy to use (requires java JDK 11 installed) and we can compute accessibility by transit + walk, transit + bike, specific transit modes, max_walking_distance, max transfers in transit legs, bike Level of Traffic Stress (LTS), probability of traveling in a time window, etc.
Results
Results for Lisbon, for a peak hour (7am wednesday) and non-peak hour (10pm sunday), with no transfers and 1 transfer max, in a 2h time-window, and max walking of 1km.
| No transfers | One transfer max |
|---|---|
 |
 |
 |
 |
- Regarding the accessibility for population to ZER:
With 1 transfer
| Peak hour | Non-peak hour |
|---|---|
| 50,1% in 30 min | 28,8% in 30 min |
| 96,9% in 45 min | 83,8% in 45 min |
| 99,7% in 60 min | 99,15% in 60 min |
With no transfers
| Peak hour | Non-peak hour |
|---|---|
| 41,5% in 30 min | 22,8% in 30 min |
| 78,9% in 45 min | 66,4% in 45 min |
| 89,7% in 60 min | 77,7% in 60 min |