This paper presents a new approach to genetic—based modeling for nonlinear time series analysis. The research is based on the concepts of evolution theory as well as natural selection, and hence is called “genetic modeling”. In order to find a predictive model from the nonlinear time series, we make use of ‘survival of the fittest’ principle of evolution. Through the process of genetic evolution, the AIC criteria are used as the performance measure, and the membership functions of the best-fitting models are the performance index of a chromosome. An empirical example shows that the genetic model can effectively find an intuitive model for nonlinear time series, especially when structure changes occur.
Relation:
Data Mining and Computational Intelligence pp 337-356 Part of the Studies in Fuzziness and Soft Computing book series (STUDFUZZ, volume 68)