Journal of the Korean Data and Information Science Society
- 제17권3호
- /
- Pages.821-830
- /
- 2006
- /
- 1598-9402(pISSN)
A Study on the Support Vector Machine Based Fuzzy Time Series Model
초록
This paper develops support vector based fuzzy linear and nonlinear regression models and applies it to forecasting the exchange rate. We use the result of Tanaka(1982, 1987) for crisp input and output. The model makes it possible to forecast the best and worst possible situation based on fewer than 50 observations. We show that the developed model is good through real data.