A NO-LINEAR HYBRID MODEL FOR MULTI-STEP-AHEAD FORECASTING OF CHAOTIC TIME-SERIES

• Main Author: ABDULKADIR, SAID JADID
• Format: Thesis
• Language: English
• Published: 2016
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://utpedia.utp.edu.my/id/eprint/21532/1/2015%20-COMPUTER%20%26%20INFORMATION%20SCIENCES%20-%20A%20NON-LINEAR%20HYBRID%20MODEL%20FOR%20MULTI-STEP-AHEAD%20FORECASTING%20OF%20CHAOTIC%20TIME-SERIES%20-%20SAID%20JADID%20ABDULKADIR.pdf; http://utpedia.utp.edu.my/id/eprint/21532/

Forecasting of chaotic time-series has increasingly become a popular and challenging subject. Many of the forecasting methods proposed in the literature are either inefficient when applied to multi 'itep-ahead forecasting of chaotic time series as they only perform one-step-ahead forecasts, or difficult to implement in terms of model complexity. The motivation to conduct the current study is to develop a more effective, easy-to-use and practical method for multi-step-ahead forecasting of chaotic time-series. Over the last decade. the main advances in forecasting are hybrid and ensemble modelling. Theoretical and empirical studies reported in the literature suggest that one of the best ways of enhancing forecasting performance is by hybrid modelling. where the models that constitutes the hybrid model function in a different manner hence capturing disparate data patterns.