Markov Bases in Algebraic Statistics
Description
Algebraic statistics is a rapidly developing field, where ideas from statistics and algebra meet and stimulate new research directions. One of the origins of algebraic statistics is the work by Diaconis and Sturmfels in 1998 on the use of Gröbner bases for constructing a connected Markov chain for performing conditional tests of a discrete exponential family. In this book we take up this topic and present a detailed summary of developments following the seminal work of Diaconis and Sturmfels.
This book is intended for statisticians with minimal backgrounds in algebra. As we ourselves learned algebraic notions through working on statistical problems and collaborating with notable algebraists, we hope that this book with many practical statistical problems is useful for statisticians to start working on the field.
Reviews / Votes
From the reviews:
"The book by Aoki, Hara, and Takemura presents a thorough introduction to Markov chain Monte Carlo tests for discrete exponential families, focusing on the concept of Markov bases. It is an authoritative and highly readable account of this field. . This text is the definitive reference on the subject, aimed principally at statisticians interested in Markov chain algorithms for sampling from discrete exponential families and its various applications . . It could also be used as a textbook for an advanced seminar on the subject." (Luis David García-Puente, Mathematical Reviews, December, 2013)More details
Other editions
Additional editions
Persons
Satoshi Aoki obtained his doctoral degree from the University of Tokyo in 2004 and is currently an associate professor in the Graduate School of Science and Engineering, Kagoshima University.
Hisayuki Hara obtained his doctoral degree from the University of Tokyo in 1999 and is currently an associate professor in the Faculty of Economics, Niigata University.
Akimichi Takemura obtained his doctoral degree from Stanford University in 1982 and is currently a professor in the Graduate School of Information Science and Technology, University of Tokyo.
Content
Exact tests for contingency tables and discrete exponential families.- Markov chain Monte Carlo methods over discrete sample space.- Toric ideals and their Gröbner bases.- Definition of Markov bases and other bases.- Structure of minimal Markov bases.- Method of distance reduction.- Symmetry of Markov bases.- Decomposable models of contingency tables.- Markov basis for no-three-factor interaction models and some other hierarchical models.- Two-way tables with structural zeros and fixed subtable sums.- Regular factorial designs with discrete response variables.- Group-wise selection models.- The set of moves connecting specific fibers.- Disclosure limitation problem and Markov basis.- Gröbner basis techniques for design of experiments.- Running Markov chain without Markov bases.- References.- Index.