Contribution of Fuzzy Systems for Time Series Analysis

A time series is a realization or sample function from a certain stochastic process. The main goals of the analysis of time series are forecasting, modeling and characterizing. Conventional time series models i.e. autoregressive (AR), moving average (MA), hybrid AR and MA (ARMA) models, assume that the time series is stationary. The other methods to model time series are soft computing techniques that include fuzzy systems, neural networks, genetic algorithms and hybrids. That techniques have been used to model the complexity of relationships in nonlinear time series because those techniques is as universal approximators that capable to approximate any real continuous function on a compact set to any degree of accuracy. As a universal approximator, fuzzy systems have capability to model nonstationary time series. Not all kinds of series data can be analyzed by conventional time series methods. Song & Chissom [19] introduced fuzzy time series as a dynamic process with linguistic values as its observations. Techniques to model fuzzy time series data are based on fuzzy systems. In this paper, we apply fuzzy model to forecast interest rate of Bank Indonesia certificate that gives better prediction accuracy than using other fuzzy time series methods and conventional statistical methods (AR and ECM).