
Author(s) 
Buja, A. (ed.) Tukey, P.A. (ed.) 

Title  Computing and graphics in statistics 
Publisher  Springer 
Year of publication  1991 
Reviewed by  Anatoly Zhigljavsky 
The volume includes 17 works devoted to different computational aspects of multivariate data analysis. These works could be divided into three groups. One of them contains 7 articles concerning statistical software. Some general philosophy of integrated statistical software systems as well as particular systems and packages are described in these articles. Another group of articles includes 5 works dealing with the visualization problem of multivariate data which is a primary problem of data analysis at recent years. More than one hundred illustrations, some of which are in color, allow the articles of this group to be attractive and bright. In spite of that there are no surveys in the book, 12 articles mentioned roughly reflect the contemporary stateofart in the field of statistical software and visualisation technique for multivariate data analysis. 5 remaining articles of the volume describe some numerical algorithms for statistical computations of different kind. Although there is a bit more of mathematics in them than in the first two groups these 5 articles do not reflect the stateofart in any part of statistics. Personally I like the article "Importance sampling for Bayesian estimation" by T. Hesterberg.
The volume is perfectly produced and well edited by two wellknown statisticians. It is a worthwhile addition to a huge literature on data analysis, it will not be lost among many other books dealing with multivariate data analysis.
Those dealing with statistical software and multivariate data analysis will constitute the main circle of the volume readers. Despite no specific fields of application like biometrics or econometrics are under discussion in the book the specialists handling specific sets of multivariate data may find a lot of valuable information helping them to analyse the data.
Purchase of the volume for the libraries would be a good investment for many companies and scientific institutions involved into multivariate data analysis.