Context-sensitive analysis of spatio-temporal trajectories within geogrphical information systems

点击翻译按钮获取中文摘要

Spatio-temporal data are growing fast, due to the collection of position from mobile phones and the generalization of portable devices (e.g. GPS - global positioning system), network traffic controllers and sensors. The purpose of this research program is to develop efficient spatio-temporal data models able to synthesize movement processes and patterns. That is needed for enhancing our knowledge of change processes beyond mobility, with applications in transportation studies, development of location-based services, urban and regional planning...Modelling change using off-the-shelf geographical information systems (GIS) or spatio-temporal databases technologies is impeded by the focus they put on atomistic pieces of data and their relative clumsiness to synthesize facts and discover trends. To overcome those limitations and to develop micro-simulation approaches for detecting spatio-temporal patterns and for simulating the evolution of thousands of actors, this research is combining several techniques: GIS, spatio-temporal databases, multi-agent multi-scale simulation, cellular automata, genetic algorithms and usual statistical analysis methods. The theoretical framework has multiple roots: Hagestrand's space-time prism, utility theory, evolution theory (genetic algorithms) and probability theory.The methodology is validated using five application cases in Quebec City: 1- weekly activity/travel scheduling made by individuals considering household structure, space and time constraints; 2- retrospective analysis of residential, professional and household trajectories of professionals; 3- multi-agent modelling of shopping behaviour; 4- modelling evolution of the urban form; 5- modelling accessibility to urban amenities using private and public transportation means. Emphasis is put on privacy issues while using personal data collected from people during surveys.