Statistical Inference: Computer Oriented Algorithmic Approach

摘要

Series for the moments of estimators and statistical functions often rely on a Taylor expansion, which is ultimately subject to modification by an expectation operator. These two phases are not of the same order of difficulty for, in general, it is the expectation of components of the series which quickly defy analytical development. The evolution of the subject in recent decades has been significantly affected by numerical analysis which is computer oriented. General characteristics are hard to define, and this paper illustrates the situations that can arise by briefly describing a few typical cases. Sign patterns and modulus patterns in components of series become discernible through extended computer methods, and this aspect poses many new developments and unsolved problems. 20 refs. (ERA citation 12:050997)