Applications of rough set theory and support vector machines in foreign exchange forecasting
碩士 === 國立暨南國際大學 === 資訊管理學系 === 95 === Recently, financial markets have been characterized by rapid liberalization and globalization, many nations have removed restrictions on exchange rates and capital flow. As a result international capital markets have been characterized by rapid integration. As m...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
2007
|
Online Access: | http://ndltd.ncl.edu.tw/handle/14566533196589846070 |
id |
ndltd-TW-095NCNU0396028 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-095NCNU03960282015-10-13T16:41:22Z http://ndltd.ncl.edu.tw/handle/14566533196589846070 Applications of rough set theory and support vector machines in foreign exchange forecasting 結合約略集合論與支援向量機於外匯漲跌幅預測之應用 Shi-Yu Chen 陳詩郁 碩士 國立暨南國際大學 資訊管理學系 95 Recently, financial markets have been characterized by rapid liberalization and globalization, many nations have removed restrictions on exchange rates and capital flow. As a result international capital markets have been characterized by rapid integration. As massive transfer of capital take place, the first impact in volatility of exchange rates and economic activities. This is why forecasting exchange rates have become an important problem, and an understanding of exchange rates trends is indispensable for both corporations and individual investors. For this reason, this study proposes using data mining to deal with the uncertain factors in foreign exchange market, forecasting the volatility of next day’s exchange rate according to the history price. Using rough set theory (RST) to select the attributes of importance and reduce the unnecessary attributes in the information table, and then derives the decision rules from decision table. And condition attributes and decision attribute are the technical indicators and the volatility of next day’s exchange rate, respective. The weak strength and the inferior discrimination are discarded after computes strength and discrimination of the decision rules. Also ascertains which rule can match test data, otherwise uses support vector machine (SVM) to train test data. While doing SVM this study also uses the particle swarm optimization (PSO) to determine parameters for SVM. This reduces the time required for choosing optimal parameters for SVM and avoids the problem of local. This hybrid model is proved to improve the forecast ability of RST by the experiment. Ping-Feng Pai 白炳豐 2007 學位論文 ; thesis 48 en_US |
collection |
NDLTD |
language |
en_US |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立暨南國際大學 === 資訊管理學系 === 95 === Recently, financial markets have been characterized by rapid liberalization and globalization, many nations have removed restrictions on exchange rates and capital flow. As a result international capital markets have been characterized by rapid integration. As massive transfer of capital take place, the first impact in volatility of exchange rates and economic activities. This is why forecasting exchange rates have become an important problem, and an understanding of exchange rates trends is indispensable for both corporations and individual investors.
For this reason, this study proposes using data mining to deal with the uncertain factors in foreign exchange market, forecasting the volatility of next day’s exchange rate according to the history price. Using rough set theory (RST) to select the attributes of importance and reduce the unnecessary attributes in the information table, and then derives the decision rules from decision table. And condition attributes and decision attribute are the technical indicators and the volatility of next day’s exchange rate, respective. The weak strength and the inferior discrimination are discarded after computes strength and discrimination of the decision rules. Also ascertains which rule can match test data, otherwise uses support vector machine (SVM) to train test data. While doing SVM this study also uses the particle swarm optimization (PSO) to determine parameters for SVM. This reduces the time required for choosing optimal parameters for SVM and avoids the problem of local. This hybrid model is proved to improve the forecast ability of RST by the experiment.
|
author2 |
Ping-Feng Pai |
author_facet |
Ping-Feng Pai Shi-Yu Chen 陳詩郁 |
author |
Shi-Yu Chen 陳詩郁 |
spellingShingle |
Shi-Yu Chen 陳詩郁 Applications of rough set theory and support vector machines in foreign exchange forecasting |
author_sort |
Shi-Yu Chen |
title |
Applications of rough set theory and support vector machines in foreign exchange forecasting |
title_short |
Applications of rough set theory and support vector machines in foreign exchange forecasting |
title_full |
Applications of rough set theory and support vector machines in foreign exchange forecasting |
title_fullStr |
Applications of rough set theory and support vector machines in foreign exchange forecasting |
title_full_unstemmed |
Applications of rough set theory and support vector machines in foreign exchange forecasting |
title_sort |
applications of rough set theory and support vector machines in foreign exchange forecasting |
publishDate |
2007 |
url |
http://ndltd.ncl.edu.tw/handle/14566533196589846070 |
work_keys_str_mv |
AT shiyuchen applicationsofroughsettheoryandsupportvectormachinesinforeignexchangeforecasting AT chénshīyù applicationsofroughsettheoryandsupportvectormachinesinforeignexchangeforecasting AT shiyuchen jiéhéyuēlüèjíhélùnyǔzhīyuánxiàngliàngjīyúwàihuìzhǎngdiēfúyùcèzhīyīngyòng AT chénshīyù jiéhéyuēlüèjíhélùnyǔzhīyuánxiàngliàngjīyúwàihuìzhǎngdiēfúyùcèzhīyīngyòng |
_version_ |
1717773782432612352 |