Bøger i Mathematics and its Applications serien

This volume presents a practical and unified approach to categorical data analysis based on the Akaike Information Criterion (AIC) and the Akaike Bayesian Information Criterion (ABIC). Conventional procedures for categorical data analysis are often inappropriate because the classical test procedures employed are too closely related to specific models. The approach described in this volume enables actual problems encountered by data analysts to be handled much more successfully. Amongst various topics explicitly dealt with are the problem of variable selection for categorical data, a Bayesian binary regression, and a nonparametric density estimator and its application to nonparametric test problems. The practical utility of the procedure developed is demonstrated by considering its application to the analysis of various data. This volume complements the volume Akaike Information Criterion Statistics which has already appeared in this series. For statisticians working in mathematics, the social, behavioural, and medical sciences, and engineering.

One service mathematics has rmdcred the 'Et moi, .... si j'avait su comment en rcvenir. human race. It has put common sense back je n'y semis point aile.' whc:rc it belongs, on the topmost shcIl next Jules Verne to the dusty callister labc:1lcd 'discarded non­ sense'. The series is divergent; thererore we may be Eric T. Bell able to do something with iL O. Hcavisidc Mathematics is a tool for thought. A highly necessary tool in a world where both feedback and non­ linearities abound. Similarly, all kinds of parts of mathematics serve as tools for other parts and for other sciences. Applying a simple rewriting rule to the quote on the right above one finds such statements as: 'One service topology has rendered mathematical physics .. .'; 'One service logic has rendered com­ puter science .. .'; 'One service category theory has rendered mathematics .. .'. All arguably true. And alI statements obtainable this way form part of the raison d'etre of this series.

The book presents important tools and techniques for treating problems in m- ern multivariate statistics in a systematic way. The ambition is to indicate new directions as well as to present the classical part of multivariate statistical analysis in this framework. The book has been written for graduate students and statis- cians who are not afraid of matrix formalism. The goal is to provide them with a powerful toolkit for their research and to give necessary background and deeper knowledge for further studies in di?erent areas of multivariate statistics. It can also be useful for researchers in applied mathematics and for people working on data analysis and data mining who can ?nd useful methods and ideas for solving their problems. Ithasbeendesignedasatextbookforatwosemestergraduatecourseonmultiva- ate statistics. Such a course has been held at the Swedish Agricultural University in 2001/02. On the other hand, it can be used as material for series of shorter courses. In fact, Chapters 1 and 2 have been used for a graduate course "e;Matrices in Statistics"e; at University of Tartu for the last few years, and Chapters 2 and 3 formed the material for the graduate course "e;Multivariate Asymptotic Statistics"e; in spring 2002. An advanced course "e;Multivariate Linear Models"e; may be based on Chapter 4. A lot of literature is available on multivariate statistical analysis written for di?- ent purposes and for people with di?erent interests, background and knowledge.