Reviews

Abstract

SignificanceVolume 6, Issue 3 p. 138-139 ReviewsFree Access Reviews First published: 24 August 2009 https://doi.org/10.1111/j.1740-9713.2009.00381.xAboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinked InRedditWechat Abstract Books reviewed in this issue. Survival and Event History Analysis: a Process Point of View O. O. Aalen, O. Borgan and H. K. Gjessing Applied Spatial Data Analysis with R Roger S. Bivand, Edzer J. Pebesma and Virgilio Gómez-Rubio Introduction to Scientific Programming and Simulation Using R O. Jones, R. Maillardet and A. Robinson Simulation and Monte Carlo, with Applications in Finance and MCMC J. S. Dagpunar Survival and Event History Analysis: a Process Point of View O. O. Aalen, O. Borgan and H. K. Gjessing; New York, Springer; xviii+540 pp.; 2008; £58.99 (hardcover); ISBN 978 0 387 20287 7 Where did the phrase “seven year itch” originate? Neither Google nor Wikipedia are helpful. I ask because Figure 1.4 of this book shows divorce rates for three cohorts of Norwegians. In all three, the rate rises quickly to a peak after about 7 years before falling back and remaining stable at a lower level. Is the same true in other countries one wonders? Aalen, Borgan and Gjessing do not help with this question, but they do help in very many other ways. Prime amongst these is in filling a niche between most books on survival analysis from standard perspectives and the counting process approach described at a more technical level in Andersen, Borgan, Gill and Keiding's 1993 masterpiece1—“the big yellow one” as it has come to be known. Our current authors take a counting approach throughout but write at a level that should be comfortable for Masters or beginning PhD students, with more breadth than Fleming and Harrington's2 accessible but older book with similar aims. Event histories consist of discrete events that occur over time. These include survival, multistate and recurrent event processes, which are illustrated through a sequence of real applications in Chapter 1, along with an overview of various topics to be explored in more depth later. Foremost among these are counting processes, the associated intensity and martingale processes and the independent censoring assumption, all of which are wonderfully described in the later parts of the opening chapter. The same is true for the stochastic process material introduced in Chapter 2. As the authors write, “Event histories unfold in time. Therefore, one would expect that tools from the theory of stochastic processes would be of considerable use in event history analysis”. So we get sufficient information on the tools and ideas of, for instance, stopping times, variation processes, transformations and stochastic integrals, all described in an easy and self-contained way, yet clearly considered as tools to help us with practical data analysis rather than as mathematical constructs of abstract interest only. A brilliant chapter on causality The statistical methodology starts in earnest in Chapter 3, which concentrates on nonparametric methods and, rather unusually, with the Nelson–Aalen estimator preceding Kaplan–Meier. This shows the emphasis on processes evolving in time—the intensity—rather than marginal properties. Nonetheless Aalen et al. do not skim these and we are given full and proper descriptions of the usual confidence intervals for survival curves, log-rank and other tests and, moving through later chapters, the Cox model, frailty and parametric approaches. The book has much more on top of these—in many parts reflecting the authors’ interests and expertise. One example is the additive regression model for event history data, which is extremely useful but often overlooked. Another is the description of nested case control techniques and a third, later in the book, is the attention given to first passage time models. This last typifies the authors’ interest in understanding the mechanisms that underlie developments over time, as does the brilliant chapter on causality. In summary, Aalen, Borgan and Gjessing have managed to write a book which is both practical and thought-provoking, wide-ranging yet focused and, above all, accessible. It will be around for a long time. Robin Henderson University of Newcastle upon Tyne Applied Spatial Data Analysis with R Roger S. Bivand, Edzer J. Pebesma and Virgilio Gómez-Rubio; New York, Springer; xiv+378 pp.; 2008; £35.99; ISBN 978 0 387 78170 9 Applied Spatial Data Analysis with R is an accessible text that demonstrates and explains the handling of spatial data using the R software platform. The authors have all been key contributors to the R spatial data analysis community, and the range of their contributions is evident from the comprehensive coverage of this work. It will appeal to those familiar with R but not spatial data, and vice versa, as well as those proficient in both and in search of a reference text. The first chapter introduces the different formats of spatial data, referring the reader to key texts along the way to supplement the brief coverage provided. The subsequent chapters make up two sections. The first presents the R packages, functions, classes and methods for handling spatial data. The second covers more specialised kinds of spatial analysis and reflects the authors’ interest in spatial point pattern analysis, geostatistics and interpolation, areal data analysis and disease mapping. The book also has a helpful companion website (http://www.asdar-book.org/), which is especially useful for downloading the data and code used in the book. Links to previous, and future, workshops enable readers to download presentations with additional worked examples from the authors. The R learning curve is a steep one, but this book provides an intuitive progression through the knowledge required for using it as a tool for comprehensive spatial data analysis. My own favourite chapters in the first section were “Visualising Spatial Data” and “Further Methods for Handling Spatial Data”. The former demonstrates R's ability to produce high quality maps of both discrete and continuous data. The numerous examples go beyond the R help files by explaining each line of code, enabling readers to pick and choose plot functions or plot annotations for use in their own work. The latter covers straightforward procedures in large commercial geographic information systems packages, such as overlay and combining attribute data, which are less intuitive in R. I found “Interpolation and Geostatistics” to be a highlight of the second section, with the chapter further strengthened by sharing a common dataset with Principles of Geographical Information Systems1, making it an ideal companion text for an in-depth understanding of the concepts covered. The topic of spatial autocorrelation is also very well covered in a subsequent chapter. A minor criticism is the lack of information on visualising 3-dimensional spatial data in R, such as digital elevation models. However, readers can find this elsewhere, for example in R Graphics2. In short, this book is highly recommended to statisticians and geographers interested in plotting and analysing spatial data. James Cheshire University College London Introduction to Scientific Programming and Simulation Using R O. Jones, R. Maillardet and A. Robinson; Chapman & Hall/CRC Press; 472 pp.; 2009; £48.99; ISBN 978 1 420 06872 6 The authors have written an excellent introduction to scientific programming with R. Their clear prose, logical structure, well-documented code and realistic examples made the book a pleasure to read. One particularly useful feature is the chapter of cases studies at the end, which not only demonstrates complete analyses but also acts as a pedagogical tool to review and integrate material introduced throughout the book. The closest competitor book is probably Rizzo's Statistical Computing with R1. Both books cover basic R usage, optimisation, numerical integration, parameter estimation and methods to generate random variables. Whereas Rizzo's book also covers boot-strapping, jackknife and permutation tests, Jones et al. provide a more extensive coverage of each topic, including more advance R programming, a greater number of motivating examples and a more detailed explanation of the R code. Both books cover the above topics at the same level but more space is given by Jones et al. to basic ideas before proceeding to more advanced material. Jones et al. also cover topics important for large projects or simulations (rather than just toy examples), such as debugging, numerical accuracy and efficiency, use of S3 and S4 classes, calling compiled code and creating packages. Another useful feature is a chapter of student projects containing six extended exercises in addition to the end of chapter exercises. Clear prose, logical structure, well-documented code—a pleasure to read I would strongly recommend this book for readers interested in using R for simulations, particularly for those new to scientific programming or R. It is also very student-friendly and would be suitable either as a course text book or for self-study. S.E. Lazic University of Cambridge Simulation and Monte Carlo, with Applications in Finance and MCMC J. S. Dagpunar; Wiley, Chichester; xiv+333 pp.; 2007; £29.95 (softback); ISBN 978 0 470 85495 2 This book assumes an understanding of common probability distributions, basic statistical ideas and a familiarity with the Maple programming environment. It is pitched at mathematics-based final year undergraduates or Masters students. Its aims are to show how values from a range of probability distributions can be generated, how the precision of estimates based on such simulations can be enhanced and to present applications in financial mathematics, operations research, reliability and Bayesian statistics. All eight main chapters end with a collection of exercises. A website, from which sample Maple procedures can be downloaded, accompanies the book. The assiduous reader of the first half will be well prepared to deploy the knowledge acquired in the applications discussed later. Methods to generate pseudo-random sequences mimicking U(0, 1) variables are described, with warnings on their possible deficiencies. Ways of generating values from other distributions are explored with useful examples that demonstrate efficient procedures; specific methods are given for a range of common distributions. Effective techniques for reducing the variance of estimators are covered. The second half begins with work on the pricing of financial derivatives showing both analytic solutions and the simulation approach. Explicit programmes are given with sample numerical calculations. Material on the Poisson process introduces methods of simulating Markov chains, renewal processes and queues, with a sample application to the efficient design of a hospital ward, and a collection of problems to test the reader's ingenuity. The chapter on Markov chain Monte Carlo assumes no prior knowledge but gives a warning to look elsewhere for a deeper account of this large field. In all these applications the author pays great attention to detail with full accounts of quite complex mathematical arguments. The last 40% of the book consists of solutions to many of the set problems and Maple procedures (with sample output) that implement the algorithms described in the text. Overall this book achieves its objectives: its contents have been road-tested on cohorts of students and enough detail is given, even for those not conversant with Maple. A good balance is struck between striving for an optimally efficient programme and for one whose logic is easy to verify. This book would be immensely useful for any practitioner seeking to learn more about this field, as well as for lecturers seeking a reliable and informative text. John Haigh Sussex University References 1Andersen, P. K., Borgan, O., Gill, R. D. and Keiding, N. (1993) Statistical Models Based on Counting Processes. New York: Springer. CrossrefGoogle Scholar 2Fleming, T. R. and Harrington, D. P. (1991) Counting Processes and Survival Analysis. Chichester: Wiley. Google Scholar 1Burrough, P. and McDonnell, R. (1998) Principles of Geographical Information Systems. Oxford: Oxford University Press. Google Scholar 2Murrell, P. (2005) R Graphics. Boca Raton, FL: Chapman & Hall/CRC. CrossrefGoogle Scholar 1Rizzo, M. (2008) Statistical Computing with R. Boca Raton, FL: Chapman & Hall/CRC Press. Google Scholar Volume6, Issue3Special Issue: Darwin 200th AnniversarySeptember 2009Pages 138-139 ReferencesRelatedInformation