火工品可靠性(或安全性)评估的Bayes方法

对火工品可靠性（或安全性）评估中通常采用的经典方法进行了剖析。介绍了数理统计方法学中近年来兴起的Ｂａｙｅｓ统计推断方法，并给出了此法在火工品二项分布、正态分布情况下，对可靠度评估的应用实例。     

基于可靠性统计分析法的制导弹药储存寿命评估

针对制导弹药价格昂贵而无法采用大规模性能试验和实弹射击方法进行可靠性研究的问题,采用可靠性统计分析法对制导弹药储存寿命进行评估。通过建立制导弹药可靠储存寿命模型,并以制导弹药火工品中的待发程控装置为研究对象,运用Bayes法对试验数据出现的＂倒挂＂现象进行了修正。依据修正后的试验数据进行计算,得到了在一定置信度下待发程控装置的储存寿命。     

长江上游地区泥石流灾害敏感性量化评价研究

泥石流敏感性分析对灾害分析评价预测具有重要作用。文章以长江上游为研究区,选择坡度、相对高差、地层岩性、年降雨量、地震烈度等5方面因素为评价因子,探讨运用GIS技术管理分析泥石流灾害信息和区域泥石流敏感性分析的方法。量化评价因子的作用分值大小,有效表达泥石流敏感性指数(DFSI),进而,结合空间分析与条件概率模型,实现了区域泥石流敏感性分析并制作了泥石流敏感性专题图。长江上游地区泥石流敏感性区域具有一定的规律性,高度敏感区主要分布在第一级阶梯与第二级阶梯的过渡地带,比较集中的区域有雅砻江中下游、安宁河、小江、普渡河、大渡河中下游、理塘河、白龙江、岷江上游、涪江上游等干支流的5～10 km范围内;中度敏感区主要分布在高度敏感区的外围约10～30 km范围内。利用历史泥石流资料验证,表明已经发生的泥石流主要分布在泥石流敏感性指数较高地区,说明敏感性分析的结果基本反映出研究区泥石流敏感性特性。     

A multilevel investigation of predictors and outcomes of shared leadership

For modern organizations, shared leadership becomes increasingly important. Knowledge on shared leadership may be limited, as past research often relies on cross-sectional data or student samples, and most studies neglect the multilevel nature of shared leadership. Our research model includes transformational leadership, trust, and organizational support as predictors of shared leadership. Furthermore, we analyze the influence of shared leadership on team performance and team creativity. In total, 160 teams with 697 employees participated in our field study. Data collection took place at three time points. To test our hypotheses, we used multilevel modeling with a Bayesian estimator. We found relationships of transformational leadership and trust with shared leadership at the team level and of transformational leadership, trust, and organizational support with shared leadership at the individual level. Furthermore, shared leadership fully mediated the effect of the three input factors on team performance and team creativity. This study contributes to the understanding of the antecedents and outcomes of shared leadership. Furthermore, the dynamic development of team processes based on an input–mediator–output model is explored. On the basis of the results, organizations can increase shared leadership behavior by focusing on transformational leadership and trust building.     

Cluster detection using Bayes factors from overparameterized cluster models

In this paper, we consider the use of a partition model to estimate regional disease rates and to detect spatial clusters. Formal inference regarding the number of partitions (or clusters) can be obtained with a reversible jump Markov chain Monte Carlo algorithm. As an alternative, we consider the ability of models with a fixed, but overly large, number of partitions to estimate regional disease rates and to provide informal inferences about the number and locations of clusters using local Bayes factors. We illustrate and compare these two approaches using data on leukemia incidence in upstate New York and data on breast cancer incidence in Wisconsin.     

公路隧道水害倾向性分级的Bayes判别法及应用

为提高公路隧道整体安全性能,保障人员安全,减少财产损失,避免公路隧道水害事故的发生,将Bayes判别理论应用于公路隧道水害倾向性判别和分级中。采用影响隧道水害发生的隧道区渗透系数、降水情况、单位涌水量、构造断裂带类型、围岩分级、隧道施工情况、防排水措施情况等7项指标作为基本判别因子;将公路隧道水害倾向性分为4个等级作为Bayes判别分析的4个正态总体。以采自典型的20组公路隧道的实测数据为训练样本,建立公路隧道水害倾向性分级的Bayes判别函数。对训练后的模型运用交叉确认估计法进行验证,然后运用该模型对6条待检验的公路隧道样本的水害倾向性进行分级。研究结果表明:构建的Bayes判别分析模型误判率极低,分级效果合理有效,可以运用于公路隧道水害倾向性的分级中,有利于公路隧道水害的预防和治理。     

基于贝叶斯理论的小麦籽粒镉铅超标风险预测

对农作物污染风险进行预测具有重大意义.基于贝叶斯定理及数据分布特征,建立了贝叶斯风险预测模型,并使用区域大田调查土壤-小麦重金属含量数据,预测小麦籽粒Cd和Pb超标风险并验证该模型的准确度.结果表明,该模型预测小麦籽粒Cd超标风险时相对偏差较小,以小麦籽粒Cd含量为变量的预测相对偏差仅为(2.66±1.87)%,以土壤DTPA-Cd含量和土壤Cd全量为变量时预测相对偏差则分别为(5.11±3.77)%和(5.88±3.87)%, 3个变量均能使预测结果与真实超标概率的平均相对偏差小于10%.预测小麦籽粒Pb超标风险时,仅小麦籽粒Pb含量的预测相对偏差小于10%.数据来源、数据分布特征和变量的选择是影响贝叶斯风险预测模型预测相对偏差的重要因素.该模型基于大田数据的先验分布,能够有效反映大田生产条件下小麦籽粒重金属与土壤因子间的相互关系,预测较准确,具有应用潜力.     

结构物震害预测的灰色贝叶斯方法

本文提出了一种结构物震害预测的新方法。该方法棋于贝叶斯理论和灰色系统理论，可以充分考虑到某类结构物震害所表现出的共性和某一特写结构物所具有的特殊性，能作出较全面的反映结构物实际状况的震害预测。     

Incorporating Collateral Data in Conservation Biology

Abstract:  Conservation biologists often need to set ecological modeling assumptions or estimate parameters from sparse data. In some cases this problem can be addressed by incorporating data from closely related species or from the same species at different sites (i.e., collateral data). Currently no structured methods exist for incorporating such information. An analogous problem in Actuarial science is to set premium rates in situations with little direct data on claim frequency or size. The rates are estimated using actuarial credibility theory, which incorporates collateral data with the direct data. actuarial credibility theory combined with the actuarial control cycle financial management process also provides an adaptive mechanism for updating assumptions. This theory may have some utility for ecologists wanting to incorporate collateral data in an adaptive management framework, a companion to approaches such as Bayesian updating. We describe the historical development of actuarial credibility theory from early ad hoc methods to empirical Bayes approaches. We explore some of the theory's strengths, such as relative simple formulae for incorporation collateral data, and we explore some of the theory's weaknesses, such as the use of the best linear approximation to the Bayes estimate. We illustrate potential applications of the theory using an example on the mortality rate of the Powerful Owl (  Ninox strenua ).     

Empirical Bayes calculations of concordance between endpoints in environmental toxicity experiments

Hierarchical models are considered for estimating the probability of agreement between two outcomes or endpoints from an environmental toxicity experiment. Emphasis is placed on generalized regression models, under which the prior mean is related to a linear combination of explanatory variables via a monotone function. This function defines the scale over which the systematic effects are modelled as additive. Specific illustration is provided for the logistic link function. The hierarchical model employs a conjugate beta prior that leads to parametric empirical Bayes estimators of the individual agreement parameters. An example from environmental carcinogenesis illustrates the methods, with motivation derived from estimation of the concordance between two species carcinogenicity outcomes. Based on a large database of carcinogenicity studies, the inter-species concordance is seen to be reasonably informative, i.e. in the range 67–84%. Stratification into pertinent potency-related sub-groups via the logistic model is seen to improve concordance estimation: for environmental stimuli at the extremes of the potency spectrum, concordance can reach well above 90%.