Researches on the Method of Bayesian Models Selections and Averages
For improving the analyses results of Bayesian model selections and averages, by which the uncertainties and risks of analyses results of natural disasters can be reduced or removed, two important problems about Bayes Factors computation, i.e. determination of parameters prior distribution and numerical integration of models, are mainly discussed and resolved firstly, then a new method of Bayes Factors computation has been proposed, finally the accuracy and effectiveness of this new method have been confirmed and compared with BIC by Monte-Carlo tests. Results show that the new method is more effective and reliable than BIC, since it can overcome the influence of many unfavorable factors by analyzing and describing the uncertainties of models parameters.
Natural disasters Bayes Factor Bayesian models selections and averages Hydrologic frequency model
Yah-Fang Sang Dong Wang Ji-Chun Wu
State Key Laboratory of Pollution Control and Resource Reuse, Department of Hydrosciences, School of State Key Laboratory of Pollution Control and Resource Reuse, Department of Hydrosciences, School of
国际会议
The 2nd International Conference on Risk Analysis and Crisis Response(第二届风险分析与危机反应国际学术研讨会)
北京
英文
420-425
2009-10-19（万方平台首次上网日期，不代表论文的发表时间）