Time Series Analysis and Its Applications

出版者:Springer

作者:Robert H. Shumway

页数:610

出版时间:2010-11-1

价格:USD 99.00

装帧:Paperback

isbn号码:9781441978646

具体描述

Time Series Analysis and Its Applications presents a balanced and comprehensive treatment of both time and frequency domain methods with accompanying theory. Numerous examples using non-trivial data illustrate solutions to problems such as evaluating pain perception experiments using magnetic resonance imaging or monitoring a nuclear test ban treaty. The book is designed to be useful as a text for graduate level students in the physical, biological and social sciences and as a graduate level text in statistics. Some parts may also serve as an undergraduate introductory course. Theory and methodology are separated to allow presentations on different levels. Material from the earlier 1988 Prentice-Hall text Applied Statistical Time Series Analysis has been updated by adding modern developments involving categorical time sries analysis and the spectral envelope, multivariate spectral methods, long memory series, nonlinear models, longitudinal data analysis, resampling techniques, ARCH models, stochastic volatility, wavelets and Monte Carlo Markov chain integration methods. These add to a classical coverage of time series regression, univariate and multivariate ARIMA models, spectral analysis and state-space models. The book is complemented by ofering accessibility, via the World Wide Web, to the data and an exploratory time series analysis program ASTSA for Windows that can be downloaded as Freeware. Robert H. Shumway is Professor of Statistics at the University of California, Davis. He is a Fellow of the American Statistical Association and a member of the Inernational Statistical Institute. He won the 1986 American Statistical Association Award for Outstanding Statistical Application and the 1992 Communicable Diseases Center Statistics Award; both awards were for joint papers on time series applications. He is the author of a previous 1988 Prentice-Hall text on applied time series analysis and is currenlty a Departmental Editor for the Journal of Forecasting. David S. Stoffer is Professor of Statistics at the University of Pittsburgh. He has made seminal contributions to the analysis of categorical time series and won the 1989 American Statistical Association Award for Outstanding Statistical Application in a joint paper analyzing categorical time series arising in infant sleep-state cycling. He is currently an Associate Editor of the Journal of Forecasting and has served as an Associate Editor for the Journal fo the American Statistical Association. --This text refers to an alternate Hardcover edition.

目录信息

Contents
1 Characteristics of Time Series 1
1.1 Introduction 1
1.2 The Nature of Time Series Data 3
1.3 Time Series Statistical Models 11
1.4 Measures of Dependence: Autocorrelation and Cross-Correlation 17
1.5 Stationary Time Series 22
1.6 Estimation of Correlation 28
1.7 Vector-Valued and Multidimensional Series 33
2 Time Series Regression and Exploratory Data Analysis 47
2.1 Introduction 47
2.2 Classical Regression in the Time Series Context 48
2.3 Exploratory Data Analysis 57
2.4 Smoothing in the Time Series Context 70
3 ARIMA Models 83
3.1 Introduction 83
3.2 Autoregressive Moving Average Models 84
3.3 Difference Equations 97
3.4 Autocorrelation and Partial Autocorrelation 102
3.5 Forecasting 108
3.6 Estimation 121
3.7 Integrated Models for Nonstationary Data 141
3.8 Building ARIMA Models 144
3.9 Multiplicative Seasonal ARIMA Models 154
4 Spectral Analysis and Filtering 173
4.1 Introduction 173
4.2 Cyclical Behavior and Periodicity 175
4.3 The Spectral Density 180
4.4 Periodogram and Discrete Fourier Transform 187
4.5 Nonparametric Spectral Estimation 196
4.6 Parametric Spectral Estimation 212
4.7 Multiple Series and Cross-Spectra 216
4.8 Linear Filters 221
4.9 Dynamic Fourier Analysis and Wavelets 228
4.10 Lagged Regression Models 242
4.11 Signal Extraction and Optimum Filtering 247
4.12 Spectral Analysis of Multidimensional Series 252
5 Additional Time Domain Topics 267
5.1 Introduction 267
5.2 Long Memory ARMA and Fractional Differencing 267
5.3 Unit Root Testing 277
5.4 GARCH Models 280
5.5 Threshold Models 289
5.6 Regression with Autocorrelated Errors 293
5.7 Lagged Regression: Transfer Function Modeling 296
5.8 Multivariate ARMAX Models 301
6 State-Space Models 319
6.1 Introduction 319
6.2 Filtering, Smoothing, and Forecasting 325
6.3 Maximum Likelihood Estimation 335
6.4 Missing Data Modifications 344
6.5 Structural Models: Signal Extraction and Forecasting 350
6.6 State-Space Models with Correlated Errors 354
6.6.1 ARMAX Models 355
6.6.2 Multivariate Regression with Autocorrelated Errors 356
6.7 Bootstrapping State-Space Models 359
6.8 Dynamic Linear Models with Switching 365
6.9 Stochastic Volatility 378
6.10 Nonlinear and Non-normal State-Space Models Using Monte Carlo Methods 387
7 Statistical Methods in the Frequency Domain 405
7.1 Introduction 405
7.2 Spectral Matrices and Likelihood Functions 409
7.3 Regression for Jointly Stationary Series 410
7.4 Regression with Deterministic Inputs 420
7.5 Random Coefficient Regression 429
7.6 Analysis of Designed Experiments 434
7.7 Discrimination and Cluster Analysis 450
7.8 Principal Components and Factor Analysis 468
7.9 The Spectral Envelope 485
Appendix A: Large Sample Theory 507
A.1 Convergence Modes 507
A.2 Central Limit Theorems 515
A.3 The Mean and Autocorrelation Functions 518
Appendix B: Time Domain Theory 527
B.1 Hilbert Spaces and the Projection Theorem 527
B.2 Causal Conditions for ARMA Models 531
B.3 Large Sample Distribution of the AR(p) Conditional Least Squares Estimators 533
B.4 The Wold Decomposition 537
Appendix C: Spectral Domain Theory 539
C.1 Spectral Representation Theorem 539
C.2 Large Sample Distribution of the DFT and Smoothed Periodogram 543
C.3 The Complex Multivariate Normal Distribution 554
Appendix R: R Supplement 559
R.1 First Things First 559
R.1.1 Included Data Sets 560
R.1.2 Included Scripts 562
R.2 Getting Started 567
R.3 Time Series Primer 571
· · · · · · (收起)

读后感

这本书简洁清晰，有充足但不多余的例子和code。对初学者合适，用作有基础的人的参考书也合适。美中不足的是3.6 estimation of ARMA parameters讲的太混乱。如果只需要对estimation算法有个概念，看analysis of financial time series 相应章节。time series初学者读这本书的话...

这本书简洁清晰，有充足但不多余的例子和code。对初学者合适，用作有基础的人的参考书也合适。美中不足的是3.6 estimation of ARMA parameters讲的太混乱。如果只需要对estimation算法有个概念，看analysis of financial time series 相应章节。time series初学者读这本书的话...

硕士期间学过时间序列分析，重点在于希尔伯特空间视角下的时间序列，需要比较强的泛函水平，学的一塌糊涂。近日因为工作愿意，需要利用时间序列分析进行一些分析建模，在quick R的主页上链接到了本书的页面，随即在互联网上下到这本书的电子版，读了一下导读和要用到的几个例子...  

硕士期间学过时间序列分析，重点在于希尔伯特空间视角下的时间序列，需要比较强的泛函水平，学的一塌糊涂。近日因为工作愿意，需要利用时间序列分析进行一些分析建模，在quick R的主页上链接到了本书的页面，随即在互联网上下到这本书的电子版，读了一下导读和要用到的几个例子...  

此书内容全面且比较新，除了传统内容（ARIMA,spectral analysis,state-space models）以外，还介绍了不少该领域中其他一些重要的topics或者新近的发展，诸如：GARCH,long-run memory process,threshold等。个人认为本书对ARIMA的介绍很好，第三章最后两节用几个例子介绍了Box-J...  

用户评价

这本书的结构组织堪称教科书级别的典范，逻辑脉络清晰得令人赞叹。它并非简单地罗列各种分析方法，而是构建了一个从基础理论到高级应用的完整知识体系。开篇对平稳性和季节性的深入剖析，为后续复杂模型的建立奠定了坚实的基础，每一个章节的过渡都衔接得天衣无缝，读者可以清晰地看到不同时间序列工具之间的内在联系和适用边界。我个人认为，书中对状态空间模型和卡尔曼滤波的讲解是全书的亮点之一。作者用了大量的篇幅，以一种极具启发性的方式，将抽象的数学概念转化为可操作的分析框架，这一点在很多同类书籍中是难以寻觅的。特别是在涉及非线性时间序列和高维数据处理的部分，作者没有采取一笔带过的方式，而是给出了扎实的理论基础和具体的计算步骤，这对于希望深入研究该领域的专业人士来说，无疑是一份宝贵的财富。阅读完这本书，我感觉自己对时间序列分析不再是停留在“会用某个软件”的层面，而是真正理解了背后的统计哲学。

不得不提的是，这本书的实操性超出了我的预期。它不仅仅是一本纯理论的参考书，更像是一本高度实用的操作手册。作者非常注重理论与软件实现的结合，书中穿插了大量使用主流统计软件（如R或Python的特定库）进行建模和诊断的详细代码片段和输出结果分析。这种“边学理论边动手实践”的模式，极大地增强了学习的代入感。我尝试着跟着书中的步骤，对自己的一个长期项目数据集进行了重构和分析，结果发现之前一直困扰我的模型选择难题，在参考了书中的诊断标准和模型选择准则后，迎刃而解。作者对于模型残差分析的重视程度值得称赞，他反复强调，一个看似完美的模型，如果残差分析不通过，其结论就站不住脚。这种严谨的态度，对于我们这些需要对分析结果负责的从业者来说，是至关重要的指导。这本书让我深刻体会到，时间序列分析的核心价值不在于拟合出复杂的方程，而在于是否能从数据中提取出可靠的、可解释的规律。

这本书的语言风格极其流畅自然，即便是初次接触时间序列分析的读者也能迅速进入状态。作者在介绍基本概念时，避免了过于生硬的数学推导，而是巧妙地结合实际案例进行阐释，使得那些看似枯燥的理论焕发出勃勃生机。例如，在讲解自回归滑动平均模型（ARMA）时，作者并未直接抛出复杂的公式，而是先通过一个真实的股票价格序列示例，引导读者思考数据点之间的依赖关系，然后再水到渠成地引入模型框架。这种教学方法极大地降低了学习曲线，让我在阅读过程中几乎没有感到任何挫败感。我尤其欣赏作者在讨论模型识别和定阶时的那种循序渐进的引导，仿佛身边有一位经验丰富的导师在耐心指导，每一步操作都有清晰的逻辑支撑和直观的图示辅助，这对于我独立进行数据分析工作有着莫大的帮助。全书的案例数据选择也十分贴合实际应用场景，从宏观经济指标到环境科学数据，覆盖面广且深度适中，让人在学习理论的同时，也潜移默化地掌握了数据处理和模型选择的艺术。

这本书在处理时间序列的“不规则性”方面展现了极高的专业水准。许多教材在处理理想化的平稳序列时游刃有余，但一旦面对现实世界中充斥着突变、异常值和不规则间隔的数据时，便显得力不从心。而这本书则系统地介绍了如何应对这些“脏数据”的挑战。无论是通过差分技巧来处理非平稳性，还是使用更先进的鲁棒性估计方法来抵抗极端观测值的影响，作者都给出了细致的讨论。我尤其对其中关于趋势分解方法的对比分析印象深刻，作者不仅介绍了经典的STL分解，还引入了基于小波分析的分解方法，并客观地对比了它们在不同类型数据上的优缺点。这种对不同分析工具的包容性和批判性审视，使得这本书的内容具有极强的生命力和适用性。它教会我的不仅仅是“如何做”，更是“为什么这样更好”，这种思维深度的提升，远比单纯掌握一两个技巧要宝贵得多。

阅读体验方面，这本书的排版和图表设计非常精良，这对于一本内容密集的统计学著作来说，实在难能可贵。数学公式的排版清晰、规范，关键概念和定理的标注也十分醒目，有效减少了阅读时的认知负荷。那些用于解释复杂概念的图示，如脉冲响应函数图、自相关函数图的视觉化展示，都做得极其直观到位，避免了读者陷入纯粹的符号迷宫中。再者，这本书的行文风格在保持学术严谨性的同时，也展现出一种恰到好处的亲切感，作者似乎总能预见到读者可能产生的困惑点，并提前给出解释和铺垫。读完这本书，我感到自己对时间序列分析的认知框架得到了极大的夯实和拓展，它不仅是一本工具书，更是一本能够引导思考、提升分析素养的深度读物，是该领域内不可多得的精品之作，强烈推荐给所有希望系统、深入学习时间序列分析的同仁们。

STAT 429

绝佳的入门教材

推荐！！

我真的是好讨厌证明啊

我觉得这辈子都不会再看这本书了