在多种无信息先验下, 将Gibbs抽样与Metropolis-Hastings算法混合的方法和重要抽样法应用于幂律过程强度函数的Bayesian预测分析, 简化Bayesian分析同时还能方便地给出强度函数及其函数的Bayes估计和区间分析. 所给预测方法不仅能预测幂律过程的未来强度, 同样适用于当前强度的预测. 在用具有精确解的数值模拟算例充分验证了文中方法的可行性、合理性和有效性之后, 将其应用于一个实例分析, 并就无信息先验中参数的选取给出一些建议.
Abstract
Under various reasonable noninformative priors, the hybrid of Gibbs sampling and Metropolis- Hastings algorithm, and importance sampling technique have been employed to Bayesian prediction of the intensity of the power law process. Bayesian analysis of the intensity of the power law process is facilitated, and then Bayes estimates and credible intervals of the intensity and functions of the intensity of the power law process can be easily obtained. The given prediction methods are exploited to predict not only the future intensity but also the current intensity. After results from a numerical simulation example with real value illustrate the feasibility, rationality and validity of presented methods, a real example is given. As for selection of noninformative priors, this paper provides some advices.
关键词
幂律过程 /
强度函数 /
Bayesian推断 /
Gibbs抽样 /
Metropolis-Hastings算法 /
重要抽样
{{custom_keyword}} /
Key words
power law process /
intensity function /
Bayesian inference /
Gibbs sampling /
Metropolis-Hastings algorithm /
importance sampling
{{custom_keyword}} /
中图分类号：
TB114.3
{{custom_clc.code}}
({{custom_clc.text}})
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] Bar-Lev S K, Lavi I, et al. Bayesian inference for the power law process[J]. Annals Institute Statistical Mathematics, 1992, 44(4): 623-639.
[2] Calabria R, Guida M, Pulcini G. A Bayes procedure for estimation of current system reliability[J]. IEEE Trans Reliability, 1992, 41(4): 616-620.
[3] 田国梁. 多台系统Weibull过程的Bayes统计推断方法[J]. 强度与环境, 1993, 1: 1-8. Tian G L. Bayes statistical inference procedures for multi-system Weibull process[J]. Structure & Environment Engineering, 1993, 1: 1-8.
[4] Sen A, Khattree R. On estimating the current intensity of failure for the power-law process[J]. Journal of Statistical Planning and Inference, 1998, 74(2): 253-272.
[5] Sen A. Bayesian estimation and prediction of the intensity of the power law process[J]. J Statist Comput Simul, 2002, 72(8): 613-631.
[6] Yu J W, Tian G L, Tang M L. Predictive analyses for nonhomogeneous poisson processes with power law using Bayesian approach[J]. Comput Stat Data Anal, 2007, 51(9): 4254-4268.
[7] 王燕萍, 吕震宙, 赵新攀. 基于MCMC的PLP未来强度的Bayesian预测分析[J]. 航空计算技术, 2010, 40(2): 1-4. Wang Y P, Lü Z Z, Zhao X P. Bayesian prediction analysis of the future intensity of the power law process based on Markov Chain Monte Carlo[J]. Aeronautical Computing Technique, 2010, 40(2): 1-4.
[8] Geman S, Geman D. Stochastic relaxation, Gibbs distribution and the Bayesian restoration of images[J]. IEEE Transaction on Pattern Analysis and Machine Intelligence, 1984, 6(6): 721-741.
[9] Gelfand A E, Smith A F M. Sampling based approaches to calculating marginal densities[J]. JASA, 1990, 85(3): 398-409.
[10] Tierney L. Markov chain for exploring posterior distributions (with discussions)[J]. Ann Statist, 1992, 22: 1701- 1762.
[11] 刘忠, 茆诗松. 分组数据的Bayes分析-Gibbs抽样方法[J]. 应用概率统计, 1997, 13(2): 211-216. Liu Z, Mao S S. Bayes analysis for grouped data-Gibbs sampling method[J]. Applied Probability Statistics, 1997, 13(2): 211-216.
[12] Chen M H, Shao Q M. Monte Carlo estimation of Bayesian credible and HPD intervals[J]. Journal of Computational Graphical Statistics, 1999, 8(1): 69-92.
[13] Kundu D, Pradhan B. Estimating the parameters of the generalized exponential distribution in presence of hybrid censoring[J]. Commun Statist Theory Methods, 2009, 38(12): 2030-2041.
[14] Bain L J. Statistical Analysis of Reliability and Life-Testing Models[M]. New York: Marcel Dekker, 1978.
[15] Yang B X. On the study of basin curve for mechanical-electronic equipments[J]. Reliability for Electronic Products and Environment Test, 1990, 4: 16-22.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
基金
国家自然科学基金(50875213); 西北工业大学基础研究基金(NPU-FFR-JC201101)
{{custom_fund}}
PDF(470 KB)
