Analysis and Prediction of Temperature Time Series Using Chaotic Approach

Abstract

Analysis and prediction of temperature time series is important because temperature changes can affect human’s health. The objectives of this study are to analyse and predict the temperature series in Jerantut, Pahang, Malaysia using chaotic approach. Modelling through chaotic approach divided into two stages; reconstruction of phase space and prediction processes. Through the reconstruction of phase space, a single scalar time series is rebuilt into a multi-dimensional phase space. This multi-dimensional phase space is used to detect the presence of chaotic dynamics through phase space plot and Cao method. The results show that the observed time series is chaotic in dynamic. Therefore, one hour ahead prediction through local mean approximation method is done. The correlation coefficient value obtained is 0.9789. The value which is approaching one reflected that the predicted time series and observed time series are close to each other. Thus, the modelling through chaotic approach is considered succeed. It is hoped that the model can help Malaysian Meteorological Department and Department of Environment Malaysia in order to improve their weather services.