Extracting structural characteristics of a nonlinear time series using genetic algorithms

Evolutionary computation is an optimisation method that can be used for extracting structural characteristics of a nonlinear time series. The work focuses on the use of a simple genetic algorithm in order to investigate if some dominant patterns of length L are good predictors for the next bit of a binary data set. That is produced by a simple transformation on a logistic system time series (raw data). Simulation results indicate that the method operates as a good feature extractor, as well as a good predictor for the L+1 bit of the dominant patterns, with prediction probability for some patterns, up to 100%. Furthermore, the method can be used on real world data and can be implemented in a parallel environment.