|
|
 |
| |
|
|
|
Comparisons of Fuzzy Time Series and Hybrid Grey Model for Non-stationary Data Forecasting |
|
|
|
PP: 409S-416S |
|
|
|
Author(s) |
|
|
|
Hsien-Lun Wong,
Jau-Min Shiu,
|
|
|
|
Abstract |
|
|
| Fuzzy time series (FTS) and Grey model (GM) have been widely applied to forecasting problem in many fields. Rather than traditional econometric model, the two models have no prerequisites for time series normality or error calibration. In this paper, the prediction performance of FTS- heuristic model, two-factor model, Markov model and GM- GM(1,1), GM-Markov, GM-Fourier is investigated. The comparison of the models is based on forecasting error of time series. The high noise data, Taiwan export amount and foreign exchange rate obtained from AEROM, Taiwan are used for models’ test. The results illustrate that the two forecasting models are appropriate for non-stationary time series. Among these models proposed, GM-Fourier residual modified model has best predictive performance. The study provides a beneficial reference of hybrid grey-based model in time series prediction. |
|
|
|
|
|
|
|
