Data Analytics for Smart, On-Demand and Connected Urban Mobility

Thanks to recent advances in big data, ubiquitous connectivity, cloud computing and mobile computing, many sharing and on-demand mobility platforms, including ride-sharing and bike-sharing systems, have proliferated in the metropolitan area worldwide. Due to their expanding social acceptance and increasing commercial successes, how to enhance their operation performance and service quality is becoming increasingly important and challenging. We have been conducting extensive mobility modeling, data analytics and algorithmic designs regarding the critical deployment issues of these urban mobility platforms.

Crowdsourcing-based bike-sharing station (BSS) reconfiguration

For crowdsourced information fusion, we develop a novel optimization framework called CBikes, (re)configuring the BSS network with crowdsourced station suggestions from online websites. Based on comprehensive real data analyses, we identify and utilize important global trip patterns to (re)configure the BSS network while balancing the local biases of individual feedbacks. Specifically, crowdsourced feedbacks, station usage history, cost and other constraints are fused into a joint optimization of BSS network configuration. We further design a semi-definite programming transformation to solve the bike station (re)placement problem efficiently and effectively.

Adaptive pricing for ride-sharing demand-supply rebalancing

For dynamic mobility-on-demand market, we develop CAPrice, a novel adaptive pricing scheme for urban MOD networks. Given accurate perception of zone-to-zone traffic flows in a city via spatio-temporal deep capsule network, CAPrice formulates a joint optimization problem by considering spatial equilibrium to balance the platform, drivers and riders/passengers with proactive pricing “signals.” Our experimental studies in New York City, Beijing and Chengdu have validated the accuracy, effectiveness and profitability (often 20% ride prediction accuracy and 30% profit improvements over the state-of-the-arts) of CAPrice in managing urban MOD networks.


  • Kang G. Shin

Research Fellows

  • Suining He


  • Suining He and Kang G. Shin, Spatio-Temporal Capsule-based Reinforcement Learning for Mobility-on-Demand Network Coordination, in World Wide Web Conference 2019 (WWW ’19), San Francisco, California, USA, May 2019.
  • Suining He and Kang G. Shin, (Re)Configuring Bike Station Network via Crowdsourced Information Fusion and Joint Optimization, in ACM MobiHoc 2018, Los Angeles, California, USA, June 2018.