ABSTRACT
Statistical ordinary least squares (OLS) regression and logistic regression (LR) models are workhorse techniques for prediction and classification, respectively. This chapter presents a comparison of genetic and statistic LR. Statistical OLS regression and LR models are workhorse techniques for prediction and classification, respectively. The engine behind any predictive model, statistical or machine learning, linear or nonlinear, is the fitness function, also known as the objective function. The OLS regression fitness function of mean squared error is easily understood. The chapter presents the GenIQ Model as the genetic LR alternative to the statistical LR model. The GenIQ Model is a flexible, any-size data method guided by the paradigm let the data define the model. Unlike the statistical regression models that use calculus as their number cruncher to yield the model equation, the GenIQ Model uses genetic programming (GP) as its number cruncher. In 1954, GP began with the evolutionary algorithms first utilized by Nils Aall Barricelli and applied to evolutionary simulations.
