New Publication in Environment and Planning B: Urban Analytics and City Science

The following paper is now available online!

Santos, J.R., Safitri, N.D., Safira, M., Varghese, V., Chikaraishi, M. (2021) Road Network Vulnerability and City-level Characteristics: A Nationwide Comparative Analysis of Japanese Cities, Environment and Planning B:Urban Analytics and City Science (Accepted).

Abstract: Climate change is making our cities more vulnerable, increasing the needs for further policy actions to make them more resilient. In particular, the transport network is critical in the first phase of disaster response. This study presents the epirical findings of a large scale, nationwide analysis of the road network vulnerability in 69 Japanese cities. We (1) identify the level of network efficiency using topological elements in its undisturbed normal state; (2) evaluate the level of network robustness under different random and targeted attack scenarios; and (3) analyze the relationship of the identified network efficiency and robustness indicators with city-level characteristics. The main findings include: (1) cities with a higher population and a higher infrastructure investment tend to be more robust under random attacks; (2) larger cities tend to be less robust to targeted attacks, presumably due to their high agglomeration of urban functions; (3) car dependency tends to make cities more vulnerable toward random attacks and less vulnerable toward targeted attacks as it indicates a weaker concentration in urban functions; and (4) a high modal share for trains tends to make cities less vulnerable toward random events as it indicates a high agglomeration of urban functions. These findings will help policymakers to prioritize their budget allocations to improve nationwide disaster resilience.

New Publication in the International Journal of Disaster Risk Reduction

The following paper is now available online!

Wu, L., Chikaraishi, M., Nguyen, T.A.H., Fujiwara, A. (2021) Analysis of post-disaster population movement by using mobile spatial statistics, International Journal of Disaster Risk Reduction, 54, 102047.

Abstract: Understanding and predicting post-disaster human movements is critical for evaluating a population’s vulnerability and resilience and developing plans for disaster evacuation, response and recovery. In this study, we attempt to analyze population movement by using mobile spatial statistics. In order to extract behavior patterns from the aggregated data, we use four different Latent Variable Analysis (LVA) methods – Independent Component Analysis (including FastICA and Spatial colored ICA), Non-negative Matrix Factorization (NMF), and Sparse Principal Component Analysis (SPCA) to analyze mobile statistics data of the disaster-affected area. The results indicate that each LVA methods has its pros and cons in extracting behavior patterns from the aggregated population. We conclude that, using multiple LVA methods and finding out the common patterns would be a robust way to understand and explain population dynamics. Finally, we argue that using mobile spatial statistics would be a feasible and practical option to estimate the dynamic change of human population after the occurrence of disasters.

New Publication in the Journal of Big Data Analytics in Transportation

The following paper is now available online!

Varghese, V. Chikaraishi, M., Urata, J. (2020) Deep Learning in Transport Studies: A Meta-Analysis on the Prediction Accuracy, Journal of Big Data Analytics in Transportation, Vol. 2, 199–220.

Deep learning methods are being increasingly applied in transport studies, while the methods require modellers to go through a try-and-error model tuning process particularly on choosing neural network structure. Moreover, the accuracy level also depends on other factors such as the type of data, sample size, region of data collection, and time of prediction. To efficiently facilitate such a model tuning process, this study attempts to summarize the relationship between the prediction accuracy of deep learning models and the factors which influence it. We conducted a comprehensive review of the literature by adopting a detailed search strategy, followed by a meta-analysis on prediction accuracy. Four separate linear mixed effects models, taking into account unobserved heterogeneities in prediction accuracy across studies, were developed to statistically test the impacts of influential factors on prediction accuracy for (a) all observations (136 studies; 2314 cases), (b) studies with MAPE, MRE, and average accuracy indicators (86 studies; 1,878 cases), (c) classification-based studies with accuracy indicator (29 studies; 220 cases), and (d) traffic forecasting studies with MAPE, MRE, and average accuracy indicators (36 studies, 991 cases). The final model includes additional factors to test the influence of sample size and time horizon of prediction variables. The findings showed that, as expected, deep learning models, particularly ones that consider spatiotemporal dependencies of transport phenomena, show better prediction accuracies compared to conventional machine learning models. We also found that, on average, the prediction accuracy is improved by 5.90% with 100 million additional data, while the accuracy is reduced by 5.28% with 100 min increase in time horizon of prediction in traffic forecasting studies. We concluded this paper with a comprehensive summary of the existing findings on the applications of deep learning to transport studies.

New publication in Travel Behavior and Society

The following paper is now available online!

Watanabe, H., Chikaraishi, M., Maruyama, T. (2021) How different are daily fluctuations and weekly rhythms in time-use behavior across urban settings? A case in two Japanese cities, Travel Behavior and Society, Vol. 22, 146-154.

For designing and evaluating the impacts of emerging mobility services, we need activity-based demand systems considering multi-week activity patterns and their variations. This study explored the inter- and intra-individual variations of time use in leisure activities during a non-working day. Specifically, the workday time-use impact on non-working-day time-use and their variations across two Japanese cities were examined in a metropolitan city (Yokohama) and a provincial city (Matsuyama). We used panel multiple discrete–continuous extreme value (MDCEV) models with five-week GPS-based data in the two cities. The results demonstrated the significant relationships between workday time-use and non-working day time-use (i.e., salient weekly rhythms) in Yokohama, but not in Matsuyama. For example, in Yokohama, (1) participants with long-time shopping on workday spent more time on recreation and less time on shopping on non-working day and (2) long-time workers on workday spent more time shopping on non-working days. Moreover, we clarified that workday time-use explained large portions of inter-individual variations in Yokohama while intra-individual variations were dominant in Matsuyama. It indicates that inhabitants in Yokohama were more affected by the time-use on workdays. In addition, activity time-use patterns in Matsuyama were more random than Yokohama since inhabitants in Matsuyama could take more flexible activity time-use patterns. Understanding such heterogeneous activity patterns will be crucial in developing and evaluating new mobility services in the future.

New publication in Transportation Research Record

The following paper is now available online!

Varghese, V., Chikaraishi, M., Kato, H. (2020) Analysis of Travel-Time Use in Crowded Trains using Discrete-Continuous Choices of Commuters in Tokyo, Japan, Transportation Research Record (Accepted).

Travel-based multitasking and the possibility to perform activities during travel are important factors that can make a transportation mode attractive. However, serious crowding in public transportation systems might adversely affect the passengers’ free choice to participate in activities during travel. This study aims to examine how crowding in public transportation systems is related to discrete-continuous choices in different types of multitasking options using a data set of 500 commuters in the Tokyo Metropolitan Area. Employing a multiple discrete-continuous extreme value model, this study investigates the relationship between crowding levels and multitasking behavior. The results show that high crowding levels, relative to being seated, have a significantly negative association with many multitasking options, which suggests the importance of seat availability. The estimation results also show that information and communication technology (ICT)-dependent leisure activities and non-ICT active activities, such as reading and talking with other passengers, have the lowest satiation and higher baseline preference constants, which indicates that they are preferred by passengers. Meanwhile, crowding levels were observed to have a significant relationship with these multitasking activities. Finally, the key findings, contributions, and policy implications of the findings are discussed.

New publication in Transport Policy (Part II)

The following paper is now available online!

Chikaraishi, M., Garg, P., Varghese, V., Yoshizoe, K., Urata, J., Shiomi, Y., Watanabe, R. (2020) On the possibility of short-term traffic prediction during disaster with machine learning approaches: An exploratory analysis, Transport Policy (Accepted).

Abstract: Since the cost and time required to finetune parameters in traditional short-term traffic prediction models such as traffic simulators are very high, the prediction models have been developed mainly for managing recurrent congestion, rather than non-recurrent congestion caused, for example, by disaster. Machine learning models are promising candidates for traffic prediction during non-recurrent congestion due to their ability to tune parameters without a-priori knowledge, while their applicability to non-recurrent conditions has rarely been explored. To fill in this gap, this study conducts an exploratory analysis on the applicability of various machine learning models during a transportation network disruption with particular focuses on their ability to predict traffic states and the interpretability of the results. The analysis is conducted by using data obtained during the massive transport network disruption which occurred in Hiroshima in July 2018 due to heavy rain and subsequent landslides. The models tested include random forest, support vector machine, XGBoost, shallow feed-forward neural network, and deep feed-forward neural network. The results indicate that random forest and XGBoost methods produced the best results in terms of prediction accuracy. On the other hand, deep neural network models produce better results in terms of the interpretability of the results, i.e., the results can be logically explained from the perspective of existing traffic flow theory. These findings indicate that the model which produces the best prediction accuracy is not always the best for practical use since it does not mimic the mechanisms of congestion occurrence.

New publication in transport Policy

The following paper has been published.

Chikaraishi, M., Khan, D., Yasuda, B., Fujiwara, A. (2020) Risk Perception and Social Acceptability of Autonomous Vehicles: A Case Study in Hiroshima, Japan, Transport Policy (In Press).

Given the impending introduction of self-driving cars to Japan within the next several years, gaining a better understanding of public opinion and risk perception of autonomous vehicles (AVs) is crucial. Though AVs have numerous potential social and economic benefits, including reduced travel time and environmental impacts, their implementation will depend on public acceptance. This study expanded on existing work by directly examining which aspects of AV use and function most affect risk perception. Participants were shown short animated video clips depicting the introduction of AVs into society at large, as well as three specific possible risk factors: system error, external interference with car controls (i.e., hacking), and the inability of the car to cope with unexpected events. Participants were then surveyed about their attitudes toward AVs and other potentially risky activities and technologies. The study established that the perceived advantages of all AV types (cars and buses, different automation levels) outweighed their perceived risks. Consistent with prior research, the two major aspects of perceived risk were dread and unfamiliarity. The results showed compared with other technologies, AV scores were neutral for dread risk but higher for unfamiliarity risk. The finding of high unfamiliarity indicates that public acceptance and perceived risks are likely to change as the public’s knowledge increases. We also found that receiving information about a potential system error indirectly reduced AV acceptability, where dread and unfamiliarity to the AV risks served as mediators. The results suggest that proper management on the diffusion of information, which includes public information campaigns, test-ride events and transparency about safety options, will likely influence the ultimate social acceptability of AVs and will be crucial towards its successful introduction on the market.

Two-day seminar on discrete choice models

We had an intensive seminar on discrete choice models. We originally planned to visit Kurashiki for the seminar, but we decided to have it on our campus due to COVID-19.

Day 1 (March 19)

13:00- Discrete choice models (logit, nested logit, and mixed logit)
13:00-15:00: Linking theory with programming code (MNL by Varun, NL by Johan, and ML by Monir, and Hybrid choice model by Haewon)
15:00-15:10: Data introduction by Monir
15:10-17:00: Exercise
17:00-18:00: Presentations and Discussions
18:00- Dinner

Day 2 (March 20)

9:00- Recursive logit models
9:00-10:30: Linking theory with code (by Ota)
10:30-11:30: Review on existing works (by Diana, Silvia, and Maya)
11:30-12:00: Data introduction by Ota
Lunch break
13:00-15:00: Exercise
15:00-16:00: Presentations and Discussions