Temporal GIS and Statistical Modelling of Personal Lifelines

Spatio-temporal data models for describing complex lifelines and trajectories of persons, as well as events that affect their evolution, are prerequisites for the statistical analysis of their relationships. Such analyses are useful to develop a better understanding of urban dynamics and social transformations. This paper develops a spatio-temporal database model for handling personal trajectories along a time line (many complementary lifelines) allowing for the statistical analysis of any pre-defined event. It combines survival analysis, Cox regression and temporal GIS. The combination of these aspects support an assessment of the likelihood of any event to occur in the life of persons at risk, after a given time delay and under some specific conditions. Our model was implemented and tested using a geo-relational approach that also supports spatial and temporal reasoning at complementary levels of abstraction. It allows the cross-analysis of several multi-dimensional lifelines to form individual trajectories. The application example is based on an historical survey of personal biographies (spatially located) of 418 professional workers living in the Quebec Metropolitan Area in 1995-96.