Page 132 - Proceeding of Atrans Young Researcher's Forum 2019_Neat

“Transportation for A Better Life:
Smart Mobility for Now and Then”

23 August 2019, Bangkok, Thailand

Artificial Intelligence Approaches for Prediction of Travel Decisions: a Case
Study of Hanoi, Vietnam
Topic number: 03 Paper Identification number: AYRF2019-018
Thi My Thanh TRUONG
1

1 Department of Transport Economics
University of Transport Technology, 54 Trieu Khuc, Thanh Xuan, Hanoi
Telephone +84-915252553, Fax. 0243-854-7695
E-mail: thanhttm@utt.edu.vn

Abstract
Analysis of travel decisions is a significant part of transport demand modeling. The selection
of travel alternatives (e.g. travel modes and destinations) highly depends on the travel behavior of
transport users, for instance, travel distance, trip cost, trip purpose, and household income. While
advances in machine learning have led to numerous powerful classifiers, their usefulness for
modeling travel decisions remains largely unexplored. This study aims at investigating the
possibility of applying Artificial Intelligence (AI) methods to predict the travel decisions through
travel behavior survey data in Hanoi, Vietnam. Firstly, travel interview survey was conducted with
the involvement of 311 transport users at different land-use types. The study secondly applied two
AI techniques namely Ensemble Decision Trees: Bagged (EDT Bagged) and Supported Vector
Machine (SVM) to predict the travel decisions of transport users. Finally, the recommendation for
a possibility of AI application on travel decisions is also proposed. The application of AI to predict
travel decisions contributes for the big data application on transport demand modeling, capturing
current trends in interdisciplinary mobility research. The results of this study are also beneficial for
transport planners and transport authorities on extending the advanced techniques to predict travel
decisions.

Keywords: Travel Decisions; Transport Demand Modeling; Machine Learning; Ensembles Decision Trees;
Supported Vector Machine

1. Introduction [2]. Therefore, proper traffic management
might have significant impact on travel decisions
In Vietnam, motorcycles and cars are the two
main modes of transport in terms of absolute volume that contribute to sustainability goal of urban
and contribution to cargo transport in the whole transport. The most important travel decisions
country, especially in urban areas. Motorcycles are include the choice of travel mode, the selection of
by far the dominant mode. By the end of 2018, route, and the preference of destination for a non-
motorcycles covered 62.7% of motorized-mode trip work trip [3]. Predicting travel decisions under
while the modal share of public transport (only buses impact factors, thereby, help transport authorities to
available) was quite small at 8.4% (Ministry of develop appropriate strategies to encourage the
Transport, 2018). Travel behavior in Vietnam is mode-shifting from motorcycles and cars to public
dominated by motorcycle-traffic-culture in which transport and active transport (e.g. walking and
the convenience of this transport mode is exploited. cycling).
With the ability to enter small alleys and to serve Discrete choice analysis is used in many
door-to-door mobility, motorcycles are excellent in contexts regarding marketing (e.g. the choice of
accessibility. They are also relatively small in size, brand), politics (e.g. the choice of president),
offering maneuvering flexibility and freedom to park economics (e.g. labor market participation),
practically nearly anywhere [1]. The dominant of medicine (e.g. response to treatment), and planning
motorcycles results in the negative impacts on traffic (e.g. the choice of residential area) and other
accident, urban congestion and environment important areas [4], [5]. In transport field, discrete

107