Generalized Latent Variable Modeling: Multilevel,Longitudinal, and Structural Equation Models
Generalized Latent Variable Modeling: Multilevel,Longitudinal, and Structural Equation Models
复制标题
DOI：
10.1198/tech.2005.s263
复制
发表时间：
2005-05
期刊：
影响因子：
2.5
通讯作者：
S. Lipovetsky
中科院分区：
文献类型：
--
作者：
S. Lipovetsky
非加性和更复杂的模型需要。第2章描述了另外五个例子，它们为后面章节中更详细的解释奠定了基础。本章还包括敏感度分析的一些属性，并对敏感度分析适用的各种设置进行了扩展。第三章给出了更详细的例子，展示了一些实际问题，在这些问题中，敏感度分析被证明是有用的。本章包括方差估计问题、莫里斯方法、蒙特卡罗滤波和估计/校准问题。第四章涉及在使用包含许多输入因素的复杂模型时，确定最重要的因素的子集。第五章介绍了与产出因子相关的方差分解方法。讨论了输入因子的正交集和非正交集。第六章介绍了适合诊断模型灵敏度分析的方法。文中详细介绍了一种称为“区域化灵敏度分析”的蒙特卡罗方法，包括贝叶斯方法和全局灵敏度分析。第7章说明了如何使用SimLab软件(作者的软件，可通过作者的网页免费下载)来实现正文中描述的许多方法。第一章介绍了如何使用SimLab分析投资者的投资组合实例。最后一章简短但有趣地介绍了敏感度分析在模型验证中的重要性，其中包括一些关于敏感度分析在确认基于数字的科学结果中的作用的引述。我发现这是一本有趣且内容丰富的书；然而，其中一些技术材料似乎过于复杂，缺乏足够的解释。给我留下的印象是，为了完全掌握敏感性分析的一些更复杂的方面，需要更多的参考资料来补充这篇文章。也许这是作者的部分目的，以刺激人们对更多信息的胃口。如果是这样的话，他们就成功了。
needed for nonadditive and more complex models. Chapter 2 describes five additional examples that provide the foundation for more detailed explanations in later chapters. The chapter also includes some of properties of sensitivity analysis and expands on the various settings in which sensitivity analysis is applicable. Chapter 3 presents the examples in greater detail by showing some practical problems in which sensitivity analysis has demonstrated usefulness. The chapter includes a variance estimation problem, the method of Morris, Monte Carlo filtering, and estimation/calibration problems. Chapter 4 concerns the identification of a subset of the most important factors when working with a complex model involving numerous input factors. Chapter 5 presents methods that illustrate the approach of variance decomposition relative to the output factor. Both orthogonal and nonorthogonal sets of input factors are discussed. Chapter 6 presents methods suitable for sensitivity analysis of diagnostic models. A Monte Carlo approach termed “regionalized sensitivity analysis” is presented in detail, and Bayesian methods and global sensitivity analysis are covered. Chapter 7 illustrates use of the SIMLAB software (the authors’ software, available for free download at through the authors’ webpage) to implement many of the methods described in the text. There are step-by-step instructions that explain how to use SIMLAB to analyze the investor’s portfolio example in Chapter 1. The last chapter is a short but entertaining presentation of the importance of sensitivity analysis in model validation that includes some quotations on the role of sensitivity analysis to confirm numerical based scientific findings. I found this to be an interesting and informative book; however, some of the technical material seemed overly complicated and lacking sufficient explanation. I was left with the impression that to fully grasp some of the more complex aspects of sensitivity analysis, additional references are needed to supplement this text. Perhaps that is part of the authors’ purpose, to whet the appetite for more information. If so, they have succeeded.