A Fuzzy-rough Based Approach for Time-series Prediction

摘要

A new time-series prediction method is proposed using rough set and fuzzy set theories. The rough set theory allows to obtain a linguistic description of time-series, whereas the fuzzy logic theory allows to generate numerical values of the time-series starting from its linguistic description. With this new approach, a reduction fuzzy rule base is generated from the history input-output data pairs to approximate a nonlinearity function. The computer simulation results for prediction Mackey-Glass chaotic time series show that this new method is effective and has good performance.