MAIN RESEARCH TOPICS

My main research interest is developing new methodologies for modeling and analyzing agents’ behavior in networks. The particular focus of applications is on (but not limited to) urban transportation networks, such as congested transport networks, last-mile supply chain networks, pedestrian activity networks, and human-centric city network designs. Please find below more details and selected outputs for each topic (though under construction). The list of publications is available here.

Modeling of Network Route Choice Behavior

#RouteChoice #DiscreteChoice #GPS #Network #MachineLearning

Route choice models predict which path a given agent takes to travel from a location to another in a network, and it has been a central topic to transportation planning/engineering and studied in various fields of research. However, modeling route choices is challenging mainly due to the complex structure of networks.
In particular, I am interested in the link-based approach to route choice modeling, which describes agents' route choice behavior in a network by the choice of links (edges) rather than paths (sequences of links), in a Markovian fashion. This approach allows us to model route choice behavior in a network without enumerating all possible paths, which is often infeasible in large networks, thereby providing a consistent estimator. Our achievements include the developments of model in extraordinary networks [1], method to learn behavior from raw GPS trajectories with measurement uncertainty [2], algorithms to capture positive network attributes within a link-based framework [3], and simultaneous modeling global and local routing mechanisms [4]. The proposed method is applicable not only to car networks but also to pedestrian networks, where agents' behavior is more uncertain and are affected by street attractiveness.

    Selected publications on this topic:
  1. Oyama, Y.,  Hato, E. (2017) A discounted recursive logit model for dynamic gridlock network analysis. Transportation Research Part C: Emerging Technologies 85: 509-527.
  2. Oyama, Y.,  Hato, E. (2018) Link-based measurement model to estimate route choice parameters in urban pedestrian networks. Transportation Research Part C: Emerging Technologies 93: 62-78.
  3. Oyama, Y. (2023) Capturing positive network attributes during the estimation of recursive logit models: A prism-based approach. Transportation Research Part C: Emerging Technologies 147, 104014.
  4. Oyama, Y. (2024) Global path preference and local response: A reward decomposition approach for network path choice analysis in the presence of visually perceived attributes. Transportation Research Part A: Policy and Practice 181: 103998.

Network Traffic Assignment

#TrafficAssignment #Equilibrium #RouteChoice #Network

Predicting traffic flows in a network is fundamental to transportation network planning and congestion management. Network traffic assignment is a problem to analyze traffic flow patterns in networks, given the travel demand between different origin-destination (OD) pairs. This problem is particularly difficult in congested networks, where agents interact with each other. It is also important to consider the stochasticity (heterogeneity) and dynamics of traffic flows.
I have been working on the developments of methodologies for network traffic assignment, including network loading algorithms for stable and efficient traffic assignment [1], analytical and stochastic traffic equilibrium assignment models [2,3], and efficient solution algorithms based on dual formulations [2].
Our new models and algorithms are tested on various types of networks, such as vehicular, railway, and pedestrian networks (and their combinations), but the application is not limited to transportation networks.

    Selected publications on this topic:
  1. Oyama, Y.,  Hato, E. (2019) Prism-based path set restriction for solving Markovian traffic assignment problem. Transportation Research Part B: Methodological 122: 528-546.
  2. Oyama, Y., Hara, Y., Akamatsu, T. (2022) Markovian traffic equilibrium assignment based on network generalized extreme value model. Transportation Research Part B: Methodological 155: 135-159.
  3. Akamatsu, T., Satsukawa, K., Oyama, Y. (2023) Global stability of day-to-day dynamics for scheduled-based Markovian transit assignment with boadring queues. arXiv preprint.

Pedestrian Activity and Walkability

#Pedestrian #RouteChoice #Perception #Walkability #StreetImage

Walking is a sustainable transportation mode, and pedestrian activities are key to designs of lively cities. However, pedestrian behavior in networks differs in many ways from other modes of transportation, and thus traditional travel behavior models cannot necessarily be applied to pedestrians directly. For example, when walking through a network, pedestrians may take a variety of paths in reaction to attractiveness of streets. Such attractiveness may be locally and visually perceived. They can also perform activities on streets more easily, necessitating modeling of sequences of moves and stays.
Focusing on such characteristics of pedestrians, I have developed new travel behavior modeling approaches. The proposed route choice model allows the evaluation of the positive impacts of street attributes on pedestrian willingness to walk, without restricting paths that pedestrians potentially take [1]. I also proposed a method for attribute-level analysis of pedestrian local responses to street attributes, which can be used to evaluate the walkability of streets [2]. Recently, by integrating computer vision techniques, I analyzed the effects of streetscape qualities on pedestrian perceptions/activities in urban areas [2,3,4]. Moreover, I have worked on the development of an integrated model of pedestrian activities in a time-space network. In this framework, given time-constraint, pedestrians choose time-space paths that integrally include choices of multiple routes, locations and durations of activities, thereby representing a continuous sequence of moves and stays [5].

    Selected publications on this topic:
  1. Oyama, Y. (2023) Capturing positive network attributes during the estimation of recursive logit models: A prism-based approach. Transportation Research Part C: Emerging Technologies 147, 104014.
  2. Oyama, Y. (2024) Global path preference and local response: A reward decomposition approach for network path choice analysis in the presence of visually perceived attributes. Transportation Research Part A: Policy and Practice 181: 103998.
  3. Oyama, Y. (2024) Spatial city image and its formative factors: A street-based neighborhood cognition analysis. Cities 149: 104898.
  4. Wang, J., Oyama, Y. (2025) Comparative analysis of pedestrian global-local route choice across different urban contexts. Available at SSRN.
  5. Oyama, Y., Hato, E. (2016) Pedestrian activity assignment problem with time-space constraint and path correlation. Journal of the City Planning Institute of Japan 51(3): 680-687 (in Japanese).

Network Design for Human-Centric Cities

#NetworkDesignProblem #TrafficAssignment #Optimization #Pedestrian

Network design is a problem to optimally plan/design the topology and system of a network based on demand, and has been mainly studied for transportation or supply chain networks. I am employing this network design concept toward sustainable and human-centric cities; more specifically, to design pedestrian-oriented city centers without causing significant negative impacts on car traffic.
Therefore, our approach is to formulate a network design problem in multi-modal networks of vehiclar and pedestrian traffic (and others). The framework is presented as a bi-level optimization problem, where the lower-level problem is multi-modal traffic prediction, and the upper-level problem is the network optimization (e.g., decisions on placement of pedestrianized streets or parking capacities). Thus, it allows the decisions explicitly considering their effects on network behavior and demand.

    Selected publications on this topic:
  1. Parady, G., Chikaraishi, M., Oyama, Y. (2025) A walker’s paradise ain’t a driver’s hell: Evaluating the causal effect of temporary road pedestrianization on traffic conditions of surrounding roads. Journal of Transport Geography 127: 104269.
  2. Imamura, K., Oyama, Y. (2025) Optimizing loading space locations for walkable city centers. Available at SSRN.

Last-Mile Delivery System Design

#Logistics #DemandManagement #DiscreteChoice #Optimization

E-commerce has grown rapidly in the last decade, as well as during the COVID-19 pandemic, and rising customer expectations of fast, cost-effective, and punctual delivery have significantly increased the demand for last-mile parcel delivery. The increasing demand for last-mile delivery lead to strict time constraints, imposing an excessive burden on city logistics. As a result, the workload of delivery service providers has been intensifying and is recognized today as one of the major social issues faced by many countries.
For this problem, I am conducting researches on (1) last-mile delivery demand management focusing on e-commerce users' delivery option choice behavior, and (2) the development and analysis of innovative and efficient delivery methods such as crowdsourced delivery.

    Selected publications on this topic:
  1. Oyama, Y., Fukuda, D., Imura N., Nishinari, K. (2024) Do people really want fast and precisely scheduled delivery? E-commerce customers’ valuations of home delivery timing. Journal of Retailing and Consumer Services 78: 103711.
  2. Akamatsu, T., Oyama, Y. (2024) A fluid-particle decomposition approach to matching market design for crowdsourced delivery systems. Transportation Research Part C: Emerging Technologies 166: 104738.
  3. Oyama, Y., Akamatsu, T. (2025) A market-based efficient matching mechanism for crowdsourced delivery systems with demand/supply elasticities. Transportation Research Part C: Emerging Technologies 174: 105110.
  4. Okazaki, R., Oyama, Y., Imura N., Nishinari, K. (2025) Evaluating choice-based demand management strategies for day-to-day home delivery planning. Research in Transportation Economics 113: 101615.