Abstract:
We are witnessing a global trend towards urbanization that creates larger and larger cities. This development results in increased demands for mobility. At the same time, it also brings with it increased congestion, increased greenhouse gas emissions, and reduced air quality.
Meanwhile, with the increasingly digitalization of vehicular transportation, notably the deployment of GPS devices, e.g., in smartphones and vehicle navigation devices, and sensors embedded in roads, massive volumes of data are generated that capture the traffic state of a road network. This data foundation and the above problems call for new data analysis techniques that enable high-resolution vehicular transportation services that contribute to addressing the problems.
In this thesis, we adopt an uncertain time series (UTS) approach to capture both uncertainty and temporal dependency. We solve the following three problems: 1) data sparseness in UTSs in a road network; 2) decision making among several UTSs at each time interval; and 3) future forecasting for UTSs with spatio-temporal correlations.
All interested parties are welcome. After the defense the department will be hosting a small reception in cluster 3.