Abstract
Traditional forecasting models are not very effective in most financial time series. To address the problem, this study proposes a novel system for financial modeling and forecasting. In the first stage, wavelet analysis transforms the input space of raw data to a time-scale feature space suitable for financial modeling and forecasting. A spectral clustering algorithm is then used to partition the feature space into several disjointed regions according to their time series dynamics. In the second stage, multiple kernel partial least square regressors ideally suited to each partitioned region are constructed for final forecasting. The proposed model outperforms neural networks, SVMs, and traditional GARCH models, significantly reducing root-mean-squared forecasting errors.
| Original language | English |
|---|---|
| Pages (from-to) | 6755-6764 |
| Number of pages | 10 |
| Journal | Applied Mathematics and Computation |
| Volume | 217 |
| Issue number | 15 |
| DOIs | |
| Publication status | Published - 2011 Apr 1 |
Fingerprint
All Science Journal Classification (ASJC) codes
- Computational Mathematics
- Applied Mathematics
Cite this
}
Integrating spectral clustering with wavelet based kernel partial least square regressions for financial modeling and forecasting. / Huang, Shian Chang.
In: Applied Mathematics and Computation, Vol. 217, No. 15, 01.04.2011, p. 6755-6764.Research output: Contribution to journal › Article
TY - JOUR
T1 - Integrating spectral clustering with wavelet based kernel partial least square regressions for financial modeling and forecasting
AU - Huang, Shian Chang
PY - 2011/4/1
Y1 - 2011/4/1
N2 - Traditional forecasting models are not very effective in most financial time series. To address the problem, this study proposes a novel system for financial modeling and forecasting. In the first stage, wavelet analysis transforms the input space of raw data to a time-scale feature space suitable for financial modeling and forecasting. A spectral clustering algorithm is then used to partition the feature space into several disjointed regions according to their time series dynamics. In the second stage, multiple kernel partial least square regressors ideally suited to each partitioned region are constructed for final forecasting. The proposed model outperforms neural networks, SVMs, and traditional GARCH models, significantly reducing root-mean-squared forecasting errors.
AB - Traditional forecasting models are not very effective in most financial time series. To address the problem, this study proposes a novel system for financial modeling and forecasting. In the first stage, wavelet analysis transforms the input space of raw data to a time-scale feature space suitable for financial modeling and forecasting. A spectral clustering algorithm is then used to partition the feature space into several disjointed regions according to their time series dynamics. In the second stage, multiple kernel partial least square regressors ideally suited to each partitioned region are constructed for final forecasting. The proposed model outperforms neural networks, SVMs, and traditional GARCH models, significantly reducing root-mean-squared forecasting errors.
UR - http://www.scopus.com/inward/record.url?scp=79952364639&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=79952364639&partnerID=8YFLogxK
U2 - 10.1016/j.amc.2011.01.096
DO - 10.1016/j.amc.2011.01.096
M3 - Article
AN - SCOPUS:79952364639
VL - 217
SP - 6755
EP - 6764
JO - Applied Mathematics and Computation
JF - Applied Mathematics and Computation
SN - 0096-3003
IS - 15
ER -