Paul R. Yarnold
Published: 2016-05-17
Total Pages: 396
Get eBook
Procedures to identify mathematical models explicitly yielding optimal (maximum accuracy) solutions for samples were widely studied in the past century, with literatures emerging in fields such as symbolic logic, operations research, mathematical programming, systems engineering, algorithms, computer science, machine intelligence, finance, transportation science, and management science. Broad-spectrum consensus among disparate experts indicates predictive accuracy is an objective function worthy of optimization. In the Optimal (?optimizing?) Data Analysis (ODA) statistical paradigm, an optimization algorithm is first utilized to identify the model that explicitly maximizes predictive accuracy for the sample, and then the resulting optimal performance is evaluated in the context of an application-specific exact statistical architecture. Discovered in 1990, the most basic ODA model was a distribution-free machine learning algorithm used to make maximum accuracy classifications of observations into one of two categories (pass or fail) on the basis of their score on an ordered attribute (test score). When the first book on ODA was written in 2004 a cornucopia of indisputable evidence had already amassed demonstrating that statistical models identified by ODA were more flexible, transparent, intuitive, accurate, parsimonious, and generalizable than competing models instead identified using an unintegrated menagerie of legacy statistical methods. Understanding of ODA methodology skyrocketed over the next decade, and 2014 produced the development of novometric theory ? the conceptual analogue of quantum mechanics for the statistical analysis of classical data. This point was selected to pause to write Maximizing Predictive Accuracy, as a means of organizing and making sense of all that has so-far been learned about ODA, through November of 2015.Researchers exploring ODA for the first time will appreciate the intellectually transparent, intuitive presentation involving minimal use of a few simple equations. Researchers using ODA in their work will appreciate the unmatched flexibility, simplicity, and accuracy of resulting statistical models ? and their generalizability across time and sample. ODA accommodates all metrics, requires no distributional assumptions, allows for analytic weighting of individual observations, explicitly maximizes predictive accuracy (overall, or normed against chance), and supports multiple methods of assessing validity.