Practical Geostatistics: Modeling and Spatial Analysis
Practical Geostatistics: Modeling and Spatial Analysis
复制标题
实用地统计学：建模和空间分析
DOI：
10.1198/tech.2002.s654
复制
发表时间：
2002
期刊：
影响因子：
2.5
通讯作者：
T. Coburn
中科院分区：
文献类型：
--
作者：
T. Coburn
这本书是由一位在应用统计学方面有经验的心理学教授撰写的，目标读者是非专业统计学家。读者被认为没有接受过任何统计学方面的培训。统计学概念以非技术性的方式向社会科学家、物理学家和其他应用研究人员在初级水平上解释。每一章都以有用的要点总结结束。这本书分为两个部分，共12章。第一部分对标准统计方法进行了口头和图形说明，包括异常值检测、可信区间、假设检验和线性回归。作者还讨论了为什么标准方法有时会极具误导性。从业者应该意识到，学生t检验只对正常人群有意义，它会导致对非正常人群的误导性结论。其中一些方面在书中进行了详细的讨论。本文简要讨论了一种称为Bootstrap的计算机密集重采样方法。然而，我推荐C.R.Chernick的书《Bootstrap方法》来介绍这种计算机密集型方法。它是一个很好的从业者指南，它有一个关于引导方法的广泛的参考书目，使它成为一个不可或缺的参考来源。这本书的第二部分由五章组成，集中介绍了较新的方法，如稳健的位置度量(修剪均值，M估计)及其推论；关联度量(皮尔逊相关性，Kendall‘s tau，Spearmann’s Rho)；稳健回归(Theil-Sen估计器，基于最小绝对距离法的回归，M-估计器，最小修剪平方)；以及非参数方法(Wilcoxon-Mann-Whitney检验，排列检验)。对于那些很少或根本不接触统计的人来说，这本书可能会让人们对统计学能提供什么有一个基本的了解。
The book, authored by a professor of psychology with experience in applied statistics, is aimed at non-professional statisticians. The reader is not assumed to have any prior training in statistics. The statistical concepts are explained at an elementary level in a nontechnical way to social scientists, physical scientists, and other applied researchers. Each chapter is closed with a useful summary of key points. The book is divided into two parts and has 12 chapters. The rst part provides a verbal and graphical explanation of standard statistical methods, including outlier detection, con dence intervals, hypothesis testing, and linear regression. The author also discusses why standard methods can be highly misleading at times. Practitioners should be aware that Student’s t-test is meaningful only for normal populations and that it leads to misleading conclusions for non-normal populations. Some of these aspects are discussed in detail in the book. The computer intensive resampling method called bootstrap is brie y discussed here. However, I recommend the book, Bootstrap Methods by C. R. Chernick for this computer intensive method. It is an excellent guide for practitioners and it has an extensive bibliography on bootstrap methods, making it an indispensable reference source. Part 2 of the book, consisting of ve chapters, concentrates on more recent methods, such as robust measures of location (trimmed mean, M -estimators) and their inferences; measures of association (Pearson’s correlation, Kendall’s tau, Spearmans’s rho); robust regression (Theil–Sen estimator, regression based on least absolute distance method, M -estimators, least trimmed squares); and non-parametric methods (Wilcoxon–Mann–Whitney test, permutation tests). The book may give a basic idea of what statistics can offer, for those with little or no exposure to statistics.