Using Machine Learning to Predict Perceptions of a Motorbike Ban in Hanoi

GIS Research UK 2023

Using Machine Learning to Predict Perceptions of a Motorbike Ban in Hanoi

Nick Malleson, et al.

Professor of Spatial Science, School of Geography, University of Leeds

These slides: www.nickmalleson.co.uk/presentations.html

This work has received funding from the British Academy under the Urban Infrastructures of Well-Being programme [grant number UWB190190].

Co-authors

Lex Comber, Kristina Bratkova, Phe Hoang Huu, Minh Kieu, Thanh Bui Quang, Hang Nguyen Thi Thuy, and Eric Wanjau

University of Leeds; University of Auckland; Vietnam National University; R&D Consultants

Full paper

Kieu, M., E. Wanjau, A. Comber, K. Bratkova, H. N. T. Thuy, T. B. Quang, P. H. Huu, N. Malleson (2023). Factors affecting perceptions in transport – A deep dive into the motorbike ban in Hanoi, Vietnam. Case Studies on Transport Policy 11: 100958. DOI: 10.1016/j.cstp.2023.100958

Context

Transport in Há Nội

City (8M people) growing faster than transport infrastructure

Motorbikes have replaced bicycles

"almost constant buzzing and honking" (Hansen, 2022)

2nd most polluted city in South East Asia in 2018 (Huu et al., 2021).

Serious issues of congestion and pollution

Motorbike Ban

Potential policy to reduce congestion & pollution

But motorbikes are "absolutely vital" (Hansen, 2022) for residents

Ban risks "mobility injustice" (Turner, 2020) or inadvertently increasing car use (Van et al., 2009)

UTM-HANOI

Urban Transport Modelling for Sustainable Well-Being in Hanoi

2+ year project funded by the British Academy (finished in Dec 2022)

Aim: new data collection and analysis/modelling to inform transport policy

Project website: https://urban-analytics.github.io/UTM-Hanoi

Bespoke travel survey

~30,000 responses

Ask about demographics, travel behaviour (main journeys), (aspirational) vehicle ownership, potential motorbike ban.

Map comparing household counts from the census to survey responses. — Map of surveyed households, versus the distribution of households in the most recent (2019) Vietnamese census.

Aims & Method

Develop ML model (XGBoost decision tree classification) to:

Predict how sentiment towards the ban might change if:

People better informed about the ban

Public transport more readily available

Find characteristics of respondents most strongly associated with positive/negative opinion of the ban (feature importance; not discussed here)

Inputs: responses to the survey

Output: probability of 'accepting' the ban

70% data used for training

Aims & Method

Cannot adjust parameters to make predictions (non-parametric)

To simulate changes:

Systematically modify testing dataset to simulate a change in one particular variable

All other variables constant

Use trained XGBoost model to make predictions on the simulated data

Compare output before/after change

Results (i)
What happens if more people become aware of the ban?

Graph shows the percentage of people who accept the ban increases
as we make more people aware of it — Awareness of the motorbike ban. Median probability of acceptance increases as a proportion of people (M) are changed from 'unaware' to 'aware'.

Results (ii)
What happens if the distance to public transport is reduced?

Change percentage: change of someone being chosen

Modifier: multiplies distance to nearest transport

As distance to public transport increases, the probability of accepting the ban decreases.

Graph showing that as distance to transport increases, probability
of accepting the ban decrases

Probability of accepting with varying change percentage (C) and modifier (M)

Summary / Conclusions

UTM-Hanoi

Large (N=~30,000) household travel survey

XGBoost to explore propensity to support a motorbike ban

Policy implications to improve acceptance of a ban:

Positive public awareness campaigns (seems obvious, but not necessarily)

Decrease distance to public transport (or, at least, don't allow it to increase

GIS Research UK 2023