An Application of GA-SVM in the FX Rate Forecast
| Autor: | Chun-Ying Huang, 黃俊穎 |
|---|---|
| Rok vydání: | 2014 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 102 Because of advances in computer network communication technology, the electronic trading market has developed rapidly in recent years, and has created a low cost and convenient new era of financial transactions, and also has enabled the international capital to circulate rapidly. However, the difference between the exchange rate regimes by the countries of the world makes the volatility of the exchange rate is more aggravated and increase the risk of exchange rate. So how to draft reactive strategies through accurate exchange rate forecast to reduce the risk arising from changes in exchange rates is an important subject of financial management. The purpose of the exchange rate forecast except offer to refer to institute future fiscal policy for the Government and for corporate to institute investment decisions and pricing and also can be used as reference of personal investment and financial management. More recently, data mining techniques has become one of the preferred prediction approaches to both academia and industry. The study utilizes data related to national exchange relevant information and the stock market index information over the period from 2009 to 2013 to explore the use of genetic algorithms (GA) to sieve factors which have impact on the exchange rate out as a predictive variable and combined with support vector machine (SVM) to formulate the forecast of USD to NTD exchange rate prediction model is to improve the accuracy of exchange rate prediction, and hope empirical results can be useful information to provide the investing public and corporate. Empirical results show that commonly used traditional BPNN forecasting accuracy rate is 99.0623%, while the GA-SVM prediction accuracy rate of 97.7232%. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|