Exchange rate forecasting using modified empirical mode decomposition and least squares support vector machine
Forecasting exchange rate requires a model that can capture the non-stationary and non-linearity of the exchange rate data. In this paper, empirical mode decomposition (EMD) is combines with least squares support vector machine (LSSVM) model in order to forecast daily USD/TWD exchange rate. EMD is u...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Published: |
International Center for Scientific Research and Studies
2016
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/71230/ https://www.scopus.com/inward/record.uri?eid=2-s2.0-85010427055&partnerID=40&md5=631643f0150ef2ece9a4bf0e24623da6 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Forecasting exchange rate requires a model that can capture the non-stationary and non-linearity of the exchange rate data. In this paper, empirical mode decomposition (EMD) is combines with least squares support vector machine (LSSVM) model in order to forecast daily USD/TWD exchange rate. EMD is used to decompose exchange rate data behaviors which are non-linear and nonstationary. LSSVM has been successfully used in non-linear regression estimation problems and pattern recognition. However, its input number selection is not based on any theories or techniques. In this proposed model, the exchange rate is decompose first by using EMD into several simple intrinsic mode oscillations called intrinsic mode function (IMF) and a residual. Permutation distribution clustering (PDC) is used to cluster the IMF and the residual into few groups according to their similarities in order to improve the LSSVM input. After that, LSSVM is used to forecast each of the groups and all the forecasted value is sum up in order to obtain the final exchange rate forecasting value where the best number of input for the LSSVM is determine by using partial autocorrelation function (PACF). The result shows that the modified EMD-LSSVM (MEMD-LSSVM) outperforms single LSSVM and hybrid model of EMD-LSSVM. |
|---|