Chaotic Time Series Prediction by Auto Fuzzy Regression Model

Since the pioneering work of Zadeh, fuzzy set theory has been applied to amyriad of areas. Song and Chissom introduced the concept of fuzzy time series andapplied some methods to the enrolments of the University of Alabama. Thereafter weapply fuzzy techniques for system identification and apply statistical techniques tomodelling system. An automatic methodology framework that combines fuzzytechniques and statistical techniques for nonparametric residual variance estimationis proposed. The methodology framework is creating regression model by usingfuzzy techniques. Identification is performed through learning from examplesmethod introduced by Wang and Mendel algorithm. Delta test residual noiseestimation is used in order to select the best subset of inputs as well as the numberof linguistic labels for the inputs. An experimental result for chaotic time seriesprediction are compared with statistical model and shows the advantages of theproposed methodology in terms of approximation accuracy, generalizationcapability and linguistic interpretability.

Keywords:

fuzzy techniques , statistical techniques , chaotic time series , prediction , regression