Linear Models and Time-Series Analysis
Description
Linear Models and Time-Series Analysis: Regression, ANOVA, ARMA and GARCH sets a strong foundation, in terms of distribution theory, for the linear model (regression and ANOVA), univariate time series analysis (ARMAX and GARCH), and some multivariate models associated primarily with modeling financial asset returns (copula-based structures and the discrete mixed normal and Laplace). It builds on the author's previous book, Fundamental Statistical Inference: A Computational Approach, which introduced the major concepts of statistical inference. Attention is explicitly paid to application and numeric computation, with examples of Matlab code throughout. The code offers a framework for discussion and illustration of numerics, and shows the mapping from theory to computation.
The topic of time series analysis is on firm footing, with numerous textbooks and research journals dedicated to it. With respect to the subject/technology, many chapters in Linear Models and Time-Series Analysis cover firmly entrenched topics (regression and ARMA). Several others are dedicated to very modern methods, as used in empirical finance, asset pricing, risk management, and portfolio optimization, in order to address the severe change in performance of many pension funds, and changes in how fund managers work.
* Covers traditional time series analysis with new guidelines
* Provides access to cutting edge topics that are at the forefront of financial econometrics and industry
* Includes latest developments and topics such as financial returns data, notably also in a multivariate context
* Written by a leading expert in time series analysis
* Extensively classroom tested
* Includes a tutorial on SAS
* Supplemented with a companion website containing numerous Matlab programs
* Solutions to most exercises are provided in the book
Linear Models and Time-Series Analysis: Regression, ANOVA, ARMA and GARCH is suitable for advanced masters students in statistics and quantitative finance, as well as doctoral students in economics and finance. It is also useful for quantitative financial practitioners in large financial institutions and smaller finance outlets.
More details
Other editions
Additional editions
Person
Marc S. Paolella is Professor of Empirical Finance at the University of Zurich, Switzerland. He is also the Editor of Econometrics and an Associate Editor of the Royal Statistical Society Journal Series. With almost 20 years of teaching experience, he is a frequent collaborator to journals and a member of many editorial boards and societies.
Content
Preface xiii
Part I Linear Models: Regression and ANOVA 1
1 The Linear Model 3
1.1 Regression, Correlation, and Causality 3
1.2 Ordinary and Generalized Least Squares 7
1.2.1 Ordinary Least Squares Estimation 7
1.2.2 Further Aspects of Regression and OLS 8
1.2.3 Generalized Least Squares 12
1.3 The Geometric Approach to Least Squares 17
1.3.1 Projection 17
1.3.2 Implementation 22
1.4 Linear Parameter Restrictions 26
1.4.1 Formulation and Estimation 27
1.4.2 Estimability and Identi¿ability 30
1.4.3 Moments and the Restricted GLS Estimator 32
1.4.4 Testing With h = 0 34
1.4.5 Testing With Nonzero h 37
1.4.6 Examples 37
1.4.7 Con¿dence Intervals 42
1.5 Alternative Residual Calculation 47
1.6 Further Topics 51
1.7 Problems 56
1.A Appendix: Derivation of the BLUS Residual Vector 60
1.B Appendix: The Recursive Residuals 64
1.C Appendix: Solutions 66
2 Fixed E¿ects ANOVA Models 77
2.1 Introduction: Fixed, Random, and Mixed E¿ects Models 77
2.2 Two Sample t-Tests for Di¿erences in Means 78
2.3 The Two Sample t-Test with Ignored Block E¿ects 84
2.4 One-Way ANOVA with Fixed E¿ects 87
2.4.1 The Model 87
2.4.2 Estimation and Testing 88
2.4.3 Determination of Sample Size 91
2.4.4 The ANOVA Table 93
2.4.5 Computing Con¿dence Intervals 97
2.4.6 A Word on Model Assumptions 103
2.5 Two-Way Balanced Fixed E¿ects ANOVA 107
2.5.1 The Model and Use of the Interaction Terms 107
2.5.2 Sums of Squares Decomposition without Interaction 108
2.5.3 Sums of Squares Decomposition with Interaction 113
2.5.4 Example and Codes 117
3 Introduction to Random and Mixed E¿ects Models 127
3.1 One-Factor Balanced Random E¿ects Model 128
3.1.1 Model and Maximum Likelihood Estimation 128
3.1.2 Distribution Theory and ANOVA Table 131
3.1.3 Point Estimation, Interval Estimation, and Signi¿cance Testing 137
3.1.4 Satterthwaite's Method 139
3.1.5 Use of SAS 142
3.1.6 Approximate Inference in the Unbalanced Case 143
3.1.6.1 Point Estimation in the Unbalanced Case 144
3.1.6.2 Interval Estimation in the Unbalanced Case 150
3.2 Crossed Random E¿ects Models 152
3.2.1 Two Factors 154
3.2.1.1 With Interaction Term 154
3.2.1.2 Without Interaction Term 157
3.2.2 Three Factors 157
3.3 Nested Random E¿ects Models 162
3.3.1 Two Factors 162
3.3.1.1 Both E¿ects Random: Model and Parameter Estimation 162
3.3.1.2 Both E¿ects Random: Exact and Approximate Con¿dence Intervals 167
3.3.1.3 Mixed Model Case 170
3.3.2 Three Factors 174
3.3.2.1 All E¿ects Random 174
3.3.2.2 Mixed: Classes Fixed 176
3.3.2.3 Mixed: Classes and Subclasses Fixed 177
3.4 Problems 177
3.A Appendix: Solutions 178
Part II Time-Series Analysis: ARMAX Processes 185
4 The AR(1) Model 187
4.1 Moments and Stationarity 188
4.2 Order of Integration and Long-Run Variance 195
4.3 Least Squares and ML Estimation 196
4.3.1 OLS Estimator of a 196
4.3.2 Likelihood Derivation I 196
4.3.3 Likelihood Derivation II 198
4.3.4 Likelihood Derivation III 198
4.3.5 Asymptotic Distribution 199
4.4 Forecasting 200
4.5 Small Sample Distribution of the OLS and ML Point Estimators 204
4.6 Alternative Point Estimators of a 208
4.6.1 Use of the Jackknife for Bias Reduction 208
4.6.2 Use of the Bootstrap for Bias Reduction 209
4.6.3 Median-Unbiased Estimator 211
4.6.4 Mean-Bias Adjusted Estimator 211
4.6.5 Mode-Adjusted Estimator 212
4.6.6 Comparison 213
4.7 Con¿dence Intervals for a 215
4.8 Problems 219
5 Regression Extensions: AR(1) Errors and Time-varying Parameters 223
5.1 The AR(1) Regression Model and the Likelihood 223
5.2 OLS Point and Interval Estimation of a 225
5.3 Testing a = 0 in the ARX(1) Model 229
5.3.1 Use of Con¿dence Intervals 229
5.3.2 The Durbin-Watson Test 229
5.3.3 Other Tests for First-order Autocorrelation 231
5.3.4 Further Details on the Durbin-Watson Test 236
5.3.4.1 The Bounds Test, and Critique of Use of p-Values 236
5.3.4.2 Limiting Power as a ¿ ±1 239
5.4 Bias-Adjusted Point Estimation 243
5.5 Unit Root Testing in the ARX(1) Model 246
5.5.1 Null is a = 1 248
5.5.2 Null is a < 1 256
5.6 Time-Varying Parameter Regression 259
5.6.1 Motivation and Introductory Remarks 260
5.6.2 The Hildreth-Houck Random Coe¿cient Model 261
5.6.3 The TVP Random Walk Model 269
5.6.3.1 Covariance Structure and Estimation 271
5.6.3.2 Testing for Parameter Constancy 274
5.6.4 Rosenberg Return to Normalcy Model 277
6 Autoregressive and Moving Average Processes 281
6.1 AR(p) Processes 281
6.1.1 Stationarity and Unit Root Processes 282
6.1.2 Moments 284
6.1.3 Estimation 287
6.1.3.1 Without Mean Term 287
6.1.3.2 Starting Values 290
6.1.3.3 With Mean Term 292
6.1.3.4 Approximate Standard Errors 293
6.2 Moving Average Processes 294
6.2.1 MA(1) Process 294
6.2.2 MA(q) Processes 299
6.3 Problems 301
6.A Appendix: Solutions 302
7 ARMA Processes 311
7.1 Basics of ARMA Models 311
7.1.1 The Model 311
7.1.2 Zero Pole Cancellation 312
7.1.3 Simulation 313
7.1.4 The ARIMA(p, d, q) Model 314
7.2 In¿nite AR and MA Representations 315
7.3 Initial Parameter Estimation 317
7.3.1 Via the In¿nite AR Representation 318
7.3.2 Via In¿nite AR and Ordinary Least Squares 318
7.4 Likelihood-Based Estimation 322
7.4.1 Covariance Structure 322
7.4.2 Point Estimation 324
7.4.3 Interval Estimation 328
7.4.4 Model Mis-speci¿cation 330
7.5 Forecasting 331
7.5.1 AR(p) Model 331
7.5.2 MA(q) and ARMA(p, q) Models 335
7.5.3 ARIMA(p, d, q) Models 339
7.6 Bias-Adjusted Point Estimation: Extension to the ARMAX(1, q) model 339
7.7 Some ARIMAX Model Extensions 343
7.7.1 Stochastic Unit Root 344
7.7.2 Threshold Autoregressive Models 346
7.7.3 Fractionally Integrated ARMA (ARFIMA) 347
7.8 Problems 349
7.A Appendix: Generalized Least Squares for ARMA Estimation 351
7.B Appendix: Multivariate AR(p) Processes and Stationarity, and General Block Toeplitz Matrix Inversion 357
8 Correlograms 359
8.1 Theoretical and Sample Autocorrelation Function 359
8.1.1 De¿nitions 359
8.1.2 Marginal Distributions 365
8.1.3 Joint Distribution 371
8.1.3.1 Support 371
8.1.3.2 Asymptotic Distribution 372
8.1.3.3 Small-Sample Joint Distribution Approximation 375
8.1.4 Conditional Distribution Approximation 381
8.2 Theoretical and Sample Partial Autocorrelation Function 384
8.2.1 Partial Correlation 384
8.2.2 Partial Autocorrelation Function 389
8.2.2.1 TPACF: First De¿nition 389
8.2.2.2 TPACF: Second De¿nition 390
8.2.2.3 Sample Partial Autocorrelation Function 392
8.3 Problems 396
8.A Appendix: Solutions 397
9 ARMA Model Identi¿cation 405
9.1 Introduction 405
9.2 Visual Correlogram Analysis 407
9.3 Signi¿cance Tests 412
9.4 Penalty Criteria 417
9.5 Use of the Conditional SACF for Sequential Testing 421
9.6 Use of the Singular Value Decomposition 436
9.7 Further Methods: Pattern Identi¿cation 439
Part III Modeling Financial Asset Returns 443
10 Univariate GARCH Modeling 445
10.1 Introduction 445
10.2 Gaussian GARCH and Estimation 450
10.2.1 Basic Properties 451
10.2.2 Integrated GARCH 452
10.2.3 Maximum Likelihood Estimation 453
10.2.4 Variance Targeting Estimator 459
10.3 Non-Gaussian ARMA-APARCH, QMLE, and Forecasting 459
10.3.1 Extending the Volatility, Distribution, and Mean Equations 459
10.3.2 Model Mis-speci¿cation and QMLE 464
10.3.3 Forecasting 467
10.4 Near-Instantaneous Estimation of NCT-APARCH(1,1) 468
10.5 S¿¿¿¿,¿¿¿¿-APARCH and Testing the IID Stable Hypothesis 473
10.6 Mixed Normal GARCH 477
10.6.1 Introduction 477
10.6.2 The MixN(k)-GARCH(r, s) Model 478
10.6.3 Parameter Estimation and Model Features 479
10.6.4 Time-Varying Weights 482
10.6.5 Markov Switching Extension 484
10.6.6 Multivariate Extensions 484
11 Risk Prediction and Portfolio Optimization 487
11.1 Value at Risk and Expected Shortfall Prediction 487
11.2 MGARCH Constructs Via Univariate GARCH 493
11.2.1 Introduction 493
11.2.2 The Gaussian CCC and DCC Models 494
11.2.3 Morana Semi-Parametric DCC Model 497
11.2.4 The COMFORT Class 499
11.2.5 Copula Constructions 503
11.3 Introducing Portfolio Optimization 504
11.3.1 Some Trivial Accounting 504
11.3.2 Markowitz and DCC 510
11.3.3 Portfolio Optimization Using Simulation 513
11.3.4 The Univariate Collapsing Method 516
11.3.5 The ES Span 521
12 Multivariate t Distributions 525
12.1 Multivariate Student's t 525
12.2 Multivariate Noncentral Student's t 530
12.3 Jones Multivariate t Distribution 534
12.4 Shaw and Lee Multivariate t Distributions 538
12.5 The Meta-Elliptical t Distribution 540
12.5.1 The FaK Distribution 541
12.5.2 The AFaK Distribution 542
12.5.3 FaK and AFaK Estimation: Direct Likelihood Optimization 546
12.5.4 FaK and AFaK Estimation: Two-Step Estimation 548
12.5.5 Sums of Margins of the AFaK 555
12.6 MEST: Marginally Endowed Student's t 556
12.6.1 SMESTI Distribution 557
12.6.2 AMESTI Distribution 558
12.6.3 MESTI Estimation 561
12.6.4 AoNm-MEST 564
12.6.5 MEST Distribution 573
12.7 Some Closing Remarks 574
12.A ES of Convolution of AFaK Margins 575
12.B Covariance Matrix for the FaK 581
13 Weighted Likelihood 587
13.1 Concept 587
13.2 Determination of Optimal Weighting 592
13.3 Density Forecasting and Backtest Over¿tting 594
13.4 Portfolio Optimization Using (A)FaK 600
14 Multivariate Mixture Distributions 611
14.1 The Mixk Nd Distribution 611
14.1.1 Density and Simulation 612
14.1.2 Motivation for Use of Mixtures 612
14.1.3 Quasi-Bayesian Estimation and Choice of Prior 614
14.1.4 Portfolio Distribution and Expected Shortfall 620
14.2 Model Diagnostics and Forecasting 623
14.2.1 Assessing Presence of a Mixture 623
14.2.2 Component Separation and Univariate Normality 625
14.2.3 Component Separation and Multivariate Normality 629
14.2.4 Mixed Normal Weighted Likelihood and Density Forecasting 631
14.2.5 Density Forecasting: Optimal Shrinkage 633
14.2.6 Moving Averages of ¿¿¿¿ 640
14.3 MCD for Robustness and Mix2Nd Estimation 645
14.4 Some Thoughts on Model Assumptions and Estimation 647
14.5 The Multivariate Laplace and Mixk Lapd Distributions 649
14.5.1 The Multivariate Laplace and EM Algorithm 650
14.5.2 The Mixk Lapd and EM Algorithm 654
14.5.3 Estimation via MCD Split and Forecasting 658
14.5.4 Estimation of Parameter b 660
14.5.5 Portfolio Distribution and Expected Shortfall 662
14.5.6 Fast Evaluation of the Bessel Function 663
Part IV Appendices 667
Appendix A Distribution of Quadratic Forms 669
A.1 Distribution and Moments 669
A.1.1 Probability Density and Cumulative Distribution Functions 669
A.1.2 Positive Integer Moments 671
A.1.3 Moment Generating Functions 673
A.2 Basic Distributional Results 677
A.3 Ratios of Quadratic Forms in Normal Variables 679
A.3.1 Calculation of the CDF 680
A.3.2 Calculation of the PDF 681
A.3.2.1 Numeric Di¿erentiation 682
A.3.2.2 Use of Geary's formula 682
A.3.2.3 Use of Pan's Formula 683
A.3.2.4 Saddlepoint Approximation 685
A.4 Problems 689
A.A Appendix: Solutions 690
Appendix B Moments of Ratios of Quadratic Forms 695
B.1 For X ~ Nn(0, 2I) and B = I 695
B.2 For X ~ N(0, S) 708
B.3 For X ~ N(¿¿¿¿, I) 713
B.4 For X ~ N(¿¿¿¿, S) 720
B.5 Useful Matrix Algebra Results 725
B.6 Saddlepoint Equivalence Result 729
Appendix C Some Useful Multivariate Distribution Theory 733
C.1 Student's t Characteristic Function 733
C.2 Sphericity and Ellipticity 739
C.2.1 Introduction 739
C.2.2 Sphericity 740
C.2.3 Ellipticity 748
C.2.4 Testing Ellipticity 768
Appendix D Introducing the SAS Programming Language 773
D.1 Introduction to SAS 774
D.1.1 Background 774
D.1.2 Working with SAS on a PC 775
D.1.3 Introduction to the Data Step and the Program Data Vector 777
D.2 Basic Data Handling 783
D.2.1 Method 1 784
D.2.2 Method 2 785
D.2.3 Method 3 786
D.2.4 Creating Data Sets from Existing Data Sets 787
D.2.5 Creating Data Sets from Procedure Output 788
D.3 Advanced Data Handling 790
D.3.1 String Input and Missing Values 790
D.3.2 Using set with first.var and last.var 791
D.3.3 Reading in Text Files 795
D.3.4 Skipping over Headers 796
D.3.5 Variable and Value Labels 796
D.4 Generating Charts, Tables, and Graphs 797
D.4.1 Simple Charting and Tables 798
D.4.2 Date and Time Formats/Informats 801
D.4.3 High Resolution Graphics 803
D.4.3.1 The GPLOT Procedure 803
D.4.3.2 The GCHART Procedure 805
D.4.4 Linear Regression and Time-Series Analysis 806
D.5 The SAS Macro Processor 809
D.5.1 Introduction 809
D.5.2 Macro Variables 810
D.5.3 Macro Programs 812
D.5.4 A Useful Example 814
D.5.4.1 Method 1 814
D.5.4.2 Method 2 816
D.6 Problems 817
D.7 Appendix: Solutions 819
Bibliography 825
Index 875
Preface
Cowards die many times before their deaths. The valiant never taste of death but once.
(William Shakespeare, Julius Caesar, Act II, Sc. 2)
The goal of this book project is to set a strong foundation, in terms of (usually small-sample) distribution theory, for the linear model (regression and ANOVA), univariate time-series analysis (ARMAX and GARCH), and some multivariate models associated primarily with modeling financial asset returns (copula-based structures and the discrete mixed normal and Laplace). The primary target audiences of this book are masters and beginning doctoral students in statistics, quantitative finance, and economics.
This book builds on the author's "Fundamental Statistical Inference: A Computational Approach", introducing the major concepts underlying statistical inference in the i.i.d. setting, and thus serves as an ideal prerequisite for this book. I hereafter denote it as book III, and likewise refer to my books on probability theory, Paolella (2006, 2007), as books I and II, respectively. For example, Listing III.4.7 refers to the Matlab code in Program Listing 4.7, chapter 4 of book III, and likewise for references to equations, examples, and pages.
As the emphasis herein is on relatively rigorous underlying distribution theory associated with a handful of core topics, as opposed to being a sweeping monograph on linear models and time series, I believe the book serves as a solid and highly useful prerequisite to larger-scope works. These include (and are highly recommended by the author), for time-series analysis, Priestley (1981), Brockwell and Davis (1991), Hamilton (1994), and Pollock (1999); for econometrics, Hayashi (2000), Pesaran (2015), and Greene (2017); for multivariate time-series analysis, Lütkepohl (2005) and Tsay (2014); for panel data methods, Wooldridge (2010), Baltagi (2013), and Pesaran (2015); for micro-econometrics, Cameron and Trivedi (2005); and, last but far from least, for quantitative risk management, McNeil et al. (2015). With respect to the linear model, numerous excellent books dedicated to the topic are mentioned below and throughout Part I.
Notably in statistics, but also in other quantitative fields that rely on statistical methodology, I believe this book serves as a strong foundation for subsequent courses in (besides more advanced courses in linear models and time-series analysis) multivariate statistical analysis, machine learning, modern inferential methods (such as those discussed in Efron and Hastie (2016), which I mention below), and also Bayesian statistical methods. As also stated in the preface to book III, the latter topic gets essentially no treatment there or in this book, the reasons being (i) to do the subject justice would require a substantial increase in the size of these already lengthy books and (ii) numerous excellent books dedicated to the Bayesian approach, in both statistics and econometrics, and at varying levels of sophistication, already exist. I believe a strong foundation in underlying distribution theory, likelihood-based inference, and prowess in computing are necessary prerequisites to appreciate Bayesian inferential methods.
The preface to book III contains a detailed discussion of my views on teaching, textbook presentation style, inclusion (or lack thereof) of end-of-chapter exercises, and the importance of computer programming literacy, all of which are applicable here and thus need not be repeated. Also, this book, like books I, II, and III, contains far more material than could be covered in a one-semester course.
This book can be nicely segmented into its three parts, with Part I (and Appendices A and B) addressing the linear (Gaussian) model and ANOVA, Part II detailing the ARMA and ARMAX univariate time-series paradigms (along with unit root testing and time-varying parameter regression models), and Part III dedicated to modern topics in (univariate and multivariate) financial time-series analysis, risk forecasting, and portfolio optimization. Noteworthy also is Appendix C on some multivariate distributional results, with Section C.1 dedicated to the characteristic function of the (univariate and multivariate) Student's distribution, and Section C.2 providing a rather detailed discussion of, and derivation of major results associated with, the class of elliptic distributions.
A perusal of the table of contents serves to illustrate the many topics covered, and I forgo a detailed discussion of the contents of each chapter.
I now list some ways of (academically) using the book.1 All suggested courses assume a strong command of calculus and probability theory at the level of book I, linear and matrix algebra, as well as the basics of moment generating and characteristic functions (Chapters 1 and 2 from book II). All courses except the first further assume a command of basic statistical inference at the level of book III. Measure theory and an understanding of the Lebesgue integral are not required for this book.
In what follows, "Core" refers to the core chapters recommended from this book, "Add" refers to additional chapters from this book to consider, and sometimes other books, depending on interest and course focus, and "Outside" refers to recommended sources to supplement the material herein with important, omitted topics.
- One-semester beginning graduate course: Introduction to Statistics and Linear Models.
- Core (not this book): Chapters 3, 5, and 10 from book II (multivariate normal, saddlepoint approximations, noncentral distributions).
Chapters 1, 2, 3 (and parts of 7 and 8) from book III.
- Core (this book):
Chapters 1, 2, and 3, and Appendix A.
- Add: Appendix D.
- Core (not this book): Chapters 3, 5, and 10 from book II (multivariate normal, saddlepoint approximations, noncentral distributions).
- One-semester course: Linear Models.
- Core (not this book): Chapters 3, 5, and 10 from book II (multivariate normal, saddlepoint approximations, noncentral distributions).
- Core (this book):
Chapters 1, 2, and 3, and Appendix A.
- Add: Chapters 4 and 5, and Appendices B and D, select chapters from Efron and Hastie (2016).
- Outside (for regression): Select chapters from Chatterjee and Hadi (2012), Graybill and Iyer (1994), Harrell, Jr. (2015), Montgomery et al. (2012).2
- Outside (for ANOVA and mixed models): Select chapters from Galwey (2014), West et al. (2015), Searle and Gruber (2017).
- Outside (additional topics, such as generalized linear models, quantile regression, etc.): Select chapters from Khuri (2010), Fahrmeir et al. (2013), Agresti (2015).
- One-semester course: Univariate Time-Series Analysis.
- Two-semester course: Time-Series Analysis.
- Core: Chapters 4, 5, 6, 7, 8, 9, 10, and 11, and Appendices A and B.
- Add: Chapters 12 and 13, and Appendix C.
- Outside (for spectral analysis, VAR, and Kalman filtering): Select chapters from Hamilton (1994), Pollock (1999), Lütkepohl (2005), Tsay (2014), Brockwell and Davis (2016).
- Outside (for econometric topics such as GMM, use of instruments, and simultaneous equations): Select chapters from Hayashi (2000), Pesaran (2015), Greene (2017).
- One-semester course: Multivariate Financial Returns Modeling and Portfolio Optimization.
- Core (not this book): Chapters 5 and 9 (univariate mixed normal, and tail estimation) from book III.
- Core: Chapters 10, 11, 12, 13, and 14, and Appendix C.
- Add: Chapter 5 (for TVP regression such as for the CAPM).
- Outside: Select chapters from Alexander (2008), Jondeau et al. (2007), Rachev et al. (2007), Tsay (2010), Tsay (2012), and Zivot (2018).3
- Mini-course on SAS.
Appendix D is on data manipulation and basic usage of the SAS system. This is admittedly an oddity, as I use Matlab throughout (as a matrix-based prototyping language) as opposed to a primarily canned-procedure package, such as SAS, SPSS, Minitab, Eviews, Stata, etc.
The appendix serves as a tutorial on the SAS system, written in a relaxed, informal way, walking the reader through numerous examples of data input, manipulation, and merging, and use of basic statistical analysis procedures. It is included as I believe SAS still has its strengths, as discussed in its opening section, and will be around for a long time. I demonstrate its use for ANOVA in Chapters 2 and 3. As with spoken languages, knowing more than one is often useful, and in this case being fluent in one of the prototyping languages, such as Matlab, R, Python, etc., and one of (if not the arguably most important) canned-routine/data processing languages, is a smart bet for aspiring data analysts and researchers.
In line with books I, II, and III, attention is explicitly paid to application and numeric computation, with examples of Matlab code throughout. The point of including code is to offer a framework for discussion and illustration of numerics, and to show the "mapping" from theory to computation, in...
