无信息先验下平稳AR(1)模型的贝叶斯推断

摘要

研究AR(1)时间序列模型在平稳条件下的贝叶斯推断理论,构造了模型自回归系数和尺度参数的无信息先验分布,推导得到了其后验分布、后验均值、众数、中位数、分位数和最大后验区间估计,最后对几组仿真数据进行了贝叶斯分析.%In this paper,the Bayesian inference of AR(1)time series under stationary condition is presented. Under a non-informative prior,the prior distributions of the regression coefficient and the scale coefficient are derived analytically. Then the posterior distribution,posterior mean,mode,median,fractile and biggest posterior interval estimation are obtained. Finally,the Bayesian analysis is performed on several groups of simulation data.