| DC 欄位 |
值 |
語言 |
| DC.contributor | 資訊管理學系在職專班 | zh_TW |
| DC.creator | 葉建宏 | zh_TW |
| DC.creator | Chien-Hung Yeh | en_US |
| dc.date.accessioned | 2021-9-2T07:39:07Z | |
| dc.date.available | 2021-9-2T07:39:07Z | |
| dc.date.issued | 2021 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=108453013 | |
| dc.contributor.department | 資訊管理學系在職專班 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 在全球化的時代，各國經濟依賴貿易的程度日漸提升，貨幣交換的匯率波動牽動著企業的利潤多寡、國家整體經濟的成長率，而美元為全球主要交易貨幣，美元匯率的趨勢對於全球各國經濟、企業或個人資產配置影響甚鉅，美國聯邦準備系統（Federal Reserve System，Fed）相當於各國央行的角色，簡稱美聯準，其主要使命包含擴大就業、穩定物價、調控長期利率等，其中以聯邦公開巿場操作委員會(Federal Open Market Committee, FOMC)為主要管理政策單位，每年召開8次會議並發布會議紀要，又稱為美聯準會議紀要，本研究目的在以美聯準會議紀要內容探求美聯準對於美國經濟的看法及各項公布數據對於美元指數的影響，以建構深度學習中的長短期記憶模型，判斷會議紀要全文內容中對美國整體經濟的看法或是採取的政策對於美元指數(US Dollar Index, USDX)的關聯。本研究在文字探勘、情緒分析的基礎上，萃取美聯準會議紀要中對於經濟有影響的重要的關鍵字詞，結合美元指數的歷史收盤價用以建構之FOMC-Explor模型，對美聯準會議紀要公布後30天的美元指數趨勢做出預測，對比單純以歷史匯價建構的長短期記憶模型(Long Short-Term Memory, LSTM)或單純以文本分析建構的長短期本研究建構的長短期記憶模型， FOMC-Explor模型對於美聯準會議紀要發布內容後對於美元指數造成的影響，能提供具有較穩定且誤差最小的美元指數趨勢預測，作為企業及個人避險與資產配置的參考依據。 | zh_TW |
| dc.description.abstract | In the era of globalization, the economy of countries dependent on international trade. Currency is the medium of international trade and volatility of the exchange rate affects the profitability of enterprises and the growth rate of the country’s overall economy. U.S. dollar is the most popular valuation currency in international trade contract. The trend of the US dollar exchange rate has a huge impact on the global economy, corporate or personal asset allocation. The Federal Reserve System (Fed) play the role as central banks in United States of America. The missions of Fed are to expand employment, stabilize the inflation and regulate the long-term interest rates, etc. The Federal Open Market Committee (FOMC) is in charge of the monetary policy and holds 8 meetings each year and releases their meeting minutes, also known as the Federal Open Market Committee minutes. The purpose of this research is to explore the Fed’s expectation of the overall economic status from the published economic indexes and FOMC meeting minutes. Based on the text mining and sentiment analysis, this research extracts the keywords from the FOMC meeting minutes and then combines historical closed US dollar index (USDX) to construct the FOMC-Explore model. To compare the USDX trend prediction result between FOMC minutes released date to one month later then, FOMC-Explore model can provide a more stable prediction result than the Long Short-Term Memory (LSTM) model which use historical USDX or FOMC minutes. FOMC-Explore model prediction can provide corporate and individual investor a valuable reference to reduce the risk from volatility of the exchange rate and asset allocation. | en_US |
| DC.subject | 長短期記憶模型 | zh_TW |
| DC.subject | 美聯準會議紀要 | zh_TW |
| DC.subject | 美元指數 | zh_TW |
| DC.subject | FOMC-Explore模型 | zh_TW |
| DC.subject | Long Short-Term Memory | en_US |
| DC.subject | FOMC Meeting Minutes | en_US |
| DC.subject | USDX | en_US |
| DC.subject | FOMC-Explore Model | en_US |
| DC.title | 以深度學習為基礎解析美聯準會議紀要 對美元指數之預測研究 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |