时间序列趋势项的贝叶斯估计

Bayesian Estimation of Time Series Model with Trend

系统地分析了AR(p)误差项的时间序列模型的数学模型及条件似然函数,并根据似然函数的统计结构构造了模型参数的共轭先验分布,研究了正态-混合Γ先验下模型的贝叶斯推断理论,包括趋势项的核估计参数及先验参数的后验分布的统计推断.

This paper systematically analyzes the mathematic model and conditional likelihood function of time series model with AR（p） error, and constructs conjugate prior distribution of parameters on the basis of the statistical structure of the function. In the condition of normal-mixed Г prior distribution, the paper studies the Bayesian inference of the model, including the statistical inference of the parameter of the kemel estimation of trend and the prior parameters of posterior distribution.