||Functional data methods with applications to traffic flow analysis
||Jeng-Min Chiou, Institute of Statistical Science, Academia Sinica
||June 14, 13:30-14:30
||N205, Zibin Building, Fudan University
||This talk presents statistical methods for analysis and prediction of traffic flow by adopting the point of view that realizations of daily traffic volume rates come from random functions. We developed functional data approaches to clustering and classification of traffic flow patterns, regression analysis for predicting traffic flow trajectories, and multiple change point detection for a sequence of functional observations. Although motivated by applications to traffic flow analysis, these methods are widely applicable to the analysis of longitudinally recorded functional data.