Hybridization of ensemble kalman filter and non-linear auto-regressive neural network for financial forecasting

• Main Authors: Abdulkadir, S.J., Yong, S.-P., Marimuthu, M., Lai, F.-W.
• Format: Article
• Published: Springer Verlag 2014
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-84915766968&doi=10.1007%2f978-3-319-13817-6_8&partnerID=40&md5=3e4e1eb3fa629c3ffe8728d89e724cfd; http://eprints.utp.edu.my/31853/
• Institution: Universiti Teknologi Petronas

Financial data is characterized as non-linear, chaotic in nature and volatile thus making the process of forecasting cumbersome. Therefore, a successful forecasting model must be able to capture longterm dependencies from the past chaotic data. In this study, a novel hybrid model, called UKF-NARX, consists of unscented kalman filter and non-linear auto-regressive network with exogenous input trained with bayesian regulation algorithm is modelled for chaotic financial forecasting. The proposed hybrid model is compared with commonly used Elman-NARX and static forecasting model employed by financial analysts. Experimental results on Bursa Malaysia KLCI data show that the proposed hybrid model outperforms the other two commonly used models. © Springer International Publishing Switzerland 2014.