Zdenek Hlávka

There can be no question, my dear Watson, of the value of exercise before breakfast. Sherlock Holmes in “The Adventure of Black Peter” The statistical analysis of multivariate data requires a variety of techniques thatareentirelydi?erentfromtheanalysisofone-dimensionaldata. Thestudy of the joint distribution of many variables in high dimensions involves matrix techniques that are not part of standard curricula. The same is true for tra- formations and computer-intensive techniques, such as projection pursuit. The purpose of this book is to provide a set of exercises and solutions to help the student become familiar with the techniques necessary to analyze high-dimensional data. It is our belief that learning to apply multivariate statistics is like studying the elements of a criminological case. To become pro?cient, students must not simply follow a standardized procedure, they must compose with creativity the parts of the puzzle in order to see the big picture. We therefore refer to Sherlock Holmes and Dr. Watson citations as typical descriptors of the analysis. Puerile as such an exercise may seem, it sharpens the faculties of observation, and teaches one where to look and what to look for.

The authors present tools and concepts of multivariate data analysis by means of exercises and their solutions. The first part is devoted to graphical techniques. The second part deals with multivariate random variables and presents the derivation of estimators and tests for various practical situations. The last part introduces a wide variety of exercises in applied multivariate data analysis. The book demonstrates the application of simple calculus and basic multivariate methods in real life situations. It contains altogether more than 250 solved exercises which can assist a university teacher in setting up a modern multivariate analysis course. All computer-based exercises are available in the R language. All data sets are included in the library SMSdata that may be downloaded via the quantlet download center www.quantlet.org. Data sets are available also via the Springer webpage. For interactive display of low-dimensional projections of a multivariate data set, we recommend GGobi.