Some Models for Discretized Series of Events
针对环境应用中常见的离散化事件序列（二元时间序列），提出基于隐马尔可夫过程的模型，并讨论似然推断、二阶性质及多序列扩展，以西欧空气污染暴露时间为例。
A discretized series of events is a binary time series that indicates whether or not events of a point process in the line occur in successive intervals. Such data are common in environmental applications. We describe a class of models for them, based on an unobserved continuous-time discrete-state Markov process, which determines the rate of a doubly stochastic Poisson process, from which the binary time series is constructed by discretization. We discuss likelihood inference for these processes and their second-order properties and extend them to multiple series. An application involves modeling the times of exposures to air pollution at a number of receptors in Western Europe.