Abstract:
|
This research developed recurrent-event change-point models to detect the time when driving risk changes significantly for novice teenager drivers. Teenage drivers who have been newly licensed are exposed to significantly higher accident risk than beyond the initial licensure period. Previous studies showed that driving risk changed around six months after licensure. One common issue is that most existing research aggregated data into pre-defined time intervals for analysis, e.g., three-month cluster, which reduced the time resolution. Furthermore, limited research has been conducted to evaluate the change in driving risk in terms of driving time. We proposed models to allow change-points to vary among drivers with latent clustering using a parametric Bayesian approach, considering that clusters over change-point might exist among the teenagers. A Dirichlet Distribution prior was used. Monte Carlo Markov Chain algorithm was developed to sample from the posterior distribution. Model selection will be conducted to choose the best number of clusters.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.