Using Regression Analysis and Computational Intelligence Approaches to Predict Gold Prices in Taiwan
| Autor: | CHANG,CHIA-WEN, 張嘉雯 |
|---|---|
| Rok vydání: | 2016 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 104 Gold is an international currency and a financial asset that acts as a hedge against risk. It can be used to stabilize national economies and prevent domestic inflation. Accordingly, gold price volatility can affect the value of a national currency. When the economy is poor, gold can be quickly converted to cash because of its high liquidity; therefore, the average investor generally chooses to invest in gold. This study used regression analyses and computational intelligence approaches to construct prediction models for gold prices in Taiwan, and compared the predictive efficacy of the various modeling methods. In the first stage of modeling, this study used multiple regression (MR), autoregressive integrated moving average (ARIMA) model, artificial neural network (ANN), support vector regression (SVR), and multivariate adaptive regression splines (MARS) to construct single-stage prediction models. In the second stage of modeling, MR and MARS were primarily used to select critical variables for constructing mixed-phase models, namely MR–ANN, MR–SVR, MR–MARS, MARS–ANN, MARS–SVR, and MARS–MR. This study compared the predictive power of the aforementioned single-phase models and mixed-phase models by using them to predict the gold prices in Taiwan. The results revealed that the prediction efficacy of ARIMA was comparatively superior to that of the other models. Additionally, regarding practical application, when fewer explanatory variables are required to be gathered, the MR–SVR or MR–MARS models present superior prediction efficacy. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|