Statistical and Inductive Inference by Minimum Message Length

The book explores the development of a method for intrinsic classification, focusing on how to determine if a population consists of distinct types or classes based on sample data. The author reflects on the collaborative process with David Boulton, highlighting their differing perspectives on Bayesian inference and classification as a form of data encoding. Their discussions led to a unified mathematical approach that revealed the connection between inference and concise data representation, which is central to the work presented.