Context and community modelling using proximity signal
We have published our research outcomes on fundamental and practical time-series and spatio-temporal data mining and machine learning techniques which can be applied on multiple types of data from sensors, Internet of Things, and social media in the domain of smart cities, smart buildings, smart parking, and intelligent transportation systems. I have been awarded nationally competitive grants and numerous industry contracts, with the total amount awarded of over than 3M in the last 5 years. Nine out of ten of my research grants are funded collaborations with government agencies or industry, with practical applications such as indoor monitoring and analytics in university and retail environments, driving behaviour recognition, road risk analysis, and passenger movement analysis in airports. In my project with local and state governments, I developed new approaches for spatio-temporal aggregation and retrieval of urban data, and clustering heterogeneous urban data.
Our research has generated context-aware and personalised models for journey planning. Our work has generated research contributions in analysis, prediction, and personalised profiling of multiple types of trajectories (e.g. human, public transports, private cars, airplanes) from a wide range of datasets from smartphones, wireless infrastructures, sensors, and Internet of Things. We have proposed a novel contour-based accessibility aware route planning algorithm for the mobility impaired, solving the first / last mile problem while traveling on multi-modal transportation network. We have also developed data-driven predictive trip planning models for taxi passengers and drivers. An accurate taxi wait time prediction model has been tested using large-scale taxi trajectory logs from New York City.