Addressing challenges in utilising smartphone-based travel survey data using choice modelling and machine learning techniques

新型冠状病毒疫情导致我们的活动和生活方式偏好发生重大变化，并引发了重新思考城市设计的需要。这需要对影响我们长期、中期和短期出行需求的因素（例如居住地点、车辆拥有量和模式转换行为）有充分的了解。该项目旨在通过开发数学模型来预测大流行病不同阶段以及大流行病后世界的流动决策，以填补这一研究空白。这些产出旨在使城市规划者和运输提供者能够在发生新的大流行病时迅速应对变化的流动性要求。发生大流行时的行为反应是复杂和多样的。因此，拟议的研究需要调查替代行为理论，以丰富传统的活动和流动模型。它还需要大量的数据进行模型估计。该项目建议将联合收割机调查数据与被动生成的智能手机应用程序位置数据相结合，以校准这些模型。拟议研究的方法论性质及其对大数据源的关注使其特别适合AQM。关于数据融合的拟议方法工作将具有运输以外的应用。

The Covid-19 pandemic has led to substantial changes in our activity and lifestyle preferences and triggered the need to rethink the design of cities. This requires sound understanding of the factors influencing our long, medium and short-term mobility needs (residential location, vehicle ownership and mode-switching behaviour for instance). The project aims to address this research gap by developing mathematical models to predict mobility decisions in different stages of the pandemic as well as in the post-pandemic world. The outputs are intended to enable urban planners and transport providers to quickly respond to changed mobility requirements in the event of a new pandemic. Behavioural responses in the event of a pandemic are complex and heterogeneous. The proposed research hence requires investigating alternative behavioural theories to enrich the traditional activity and mobility models. It also requires extensive data for model estimation. The project proposes to combine survey data with passively generated smart-phone app location data to calibrate these models. The methodological nature of the proposed research and its focus on big data sources makes it particularly suited for the AQM. The proposed methodological work on data fusion will have applications beyond transport.