ACADSTAFF UGM
| Title | : | The development of an optimization procedure in WRBNN for time series forecasting |
| Author | : |
RUKUN SANTOSO (1) Prof. Drs. Subanar, Ph.D. (2) Prof. Dr.rer.nat. Dedi Rosadi, S.Si., M.Sc. (3) Suhartono (4) |
| Date | : | 1 2016 |
| Keyword | : | DWT, MODWT, radial basis, time series, wavelet, WRBNN DWT, MODWT, radial basis, time series, wavelet, WRBNN |
| Abstract | : | A forecasting procedure based on a wavelet radial basis neural network is proposed in this paper. The MODWT result becomes an input of the model. The smooth part constructs the main pattern of forecasting model. Meanwhile the detail parts construct the fluctuation rhythm or disturbances. The model considers that each of the transformation level contribute to the forecasting result independently. The nonlinearity properties included in the MODWT result is controlled by radial basis functions. The LM test is used to explore the number of wavelet coefficient clusters in every transformation level. The membership of cluster is determined by the k-means method. The least square method (OLS or NLS) can be used to estimate the parameters of model. |
| Group of Knowledge | : | Statistik |
| Original Language | : | English |
| Level | : | Internasional |
| Status | : |
Published
|