研究生: |
許棋堯 Hsu, Chi-Yao |
---|---|
論文名稱: |
基於長短記憶神經網路(LSTM)建構黃金價格預測模型 The Forecasting Model of Gold Price Based on Long Short Term Memory Network |
指導教授: |
張焯然
Chang, Jow-Ran |
口試委員: |
蔡璧徽
Tsai, Bi-Huei 劉剛 Liu, Kang |
學位類別: |
碩士 Master |
系所名稱: |
科技管理學院 - 財務金融 Master Program of Finance and Banking |
論文出版年: | 2020 |
畢業學年度: | 108 |
語文別: | 中文 |
論文頁數: | 22 |
中文關鍵詞: | 黃金價格 、深度學習 、LSTM |
外文關鍵詞: | Gold price, Deep learning, LSTM |
相關次數: | 點閱:2 下載:0 |
分享至: |
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
本研究主要以國際黃金價格為研究標的,在傳統上作時間序列資料的預測主要是依靠AR、MA及ARMA等模型,找出具有相關性的因子,做出相對應的預測模型。2016年後隨著各種機器學習及人工智能普遍運用在各領域,像是深度學習也開始運用在金融相關研究上。因使本研究選用長短期記憶神經網路「Long Short Term Memory Network」作為預測黃金價格的模型。資料範圍從2010年起至2019年,其中包含了每日開盤價、收盤價、最高價和最低價,利用10年的資料進行類神經網路的訓練,建構預測模型。實證結果以批次訓練筆數(Batch_size)=32,學習率=0.001,訓練次數epoch=1500,神經元個數=300,刪除比率dropout rate=0.2,可以得到MSE=0.036372,表示所建構的LSTM預測模型具有一定的預測能力。
The main research goal of this thesis is to predict international gold price. Through AR, MA and ARMA models relevant factors of international gold price can be found and produce prediction models for the gold price. After 2016, as various machine learning and artificial intelligence are commonly used in various fields, such as deep learning has also begun to be applied to financial-related research. Therefore, in this study, the Long Short Term Memory Network was selected as the model for predicting the gold price. The data from 2010 to 2019 are collected, which includes daily opening price, closing price, highest price and lowest price. The 10 years of data are applied to train the neural network and construct a prediction model. With Batch size=32, learning rate=0.001, training epoch=1500, number of neurons=300 and dropout rate=0.2, we can obtain MSE=0.036372, which represents the resulting LSTM prediction model has certain prediction ability.
參考文獻
中文部分
1.郭富城(2015),「後金融海嘯影響黃金價格之總體經濟變數之研究」,碩士論文,南華大學企業管理學系管理科學研究所。
2.陳國玄(2004),「人工神經網路與統計方法應用於台灣上市電子類股價指數預測與分類之研究」,碩士論文,國立成功大學統計學研究所。
3.鄭余建(2018) ,「以總體經濟變數預測上海A股指數-長短期記憶網路之應用」,碩士論文,天主教輔仁大學金融與國際企業系金融所。
4.謝昌儒(2016) ,「機構與個別投資人情緒指標對黃金價格之影響」,碩士論文,中央大學財務金融學系研究所。
5.
英文部分
1.Diederik K. and Jimmy B.(2014), “Adam: A Method for Stochastic Optimization”
2.Frank R. (1957), ” The Perceptron: A Probabilistic Model for Information Storage and Organization in The Brain”
3.Sepp Hochreiter and Jürgen Schmidhuber (1997),”Long Short-Term Memory.”
4.Kim, M.H. and Dilts, D.A. (2011), “The Relationship of the value of the Dollar, and the Prices of Gold and Oil: A Tale of Asset Risk”, Economic Bulletin, Vol.31, pp.1-11.
5.Kyunghyun C.(2014), “Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation”
6.Samanta, S.K. and Zadeh, A.H.M.(2012), “Co-movements of Oil, Gold, the US Dollar, and Stocks”, Modern Economy, Vol.3(1).
7.Warren M. and Walter P. (1943), “ A Logical Calculus of the Ideas Immanent in Nervous Activity.”
8.Zheng, J., Xu, C., Zhang, Z., and Li, X. (2017), “Electric Load Forecasting in Smart Grids Using Long-Short-Memory based Recurrent Neural Network”, Annual Conference of Information Sciences and Systems (CISS), pp.88-95.
網路部分
1.數位國家創新經濟發展方案,行政院,https://www.ey.gov.tw/Page/5A8A0CB5B41DA11E/f4d3319a-e2d7-4a8b-8b55-26c936804b5b