Data Fitting Based on Genetic Programming and Least Square Method

Abstract

Traditional data fitting techniques usually require estimating basis function and they are specific for different application areas. Based on dynamic characteristics of genetic programming, a two-phase data fitting algorithm is proposed. In this algorithm, genetic programming is used to optimize model structure and Least Square method is applied to estimate parameters. Proposed algorithm is tested for different types of data fitting. This high efficiency and accuracy algorithm can be applied in different areas.