本論文主要的目的在提出一個以灰關聯度改良的SOM對數列做聚類分析，再觀察聚類結果，歸納出模糊規則(Fuzzy rules)做誤差修正。並且實際使用這個新演算法套用到幾個不同的市場來檢驗股價預測的準確程度。
灰色系統理論的特性是在”部分已知、部分未知”的情形下,提取有價值的資訊。而灰關聯分析的基本概念是根據數列曲線的幾何形狀相似程度來判斷其關聯程度. 因此, 本論文嘗試以求取灰關聯度的方法來取代一般SOM以最小歐基里德距離或最大內積法找出得勝者神經元. 數列中灰關聯度最大的為得勝者神經元，找出得勝者神經元之後，調整鄰近神經元等後續步驟就依照原來SOM的流程處理。
數列的組成挑選收盤價、移動平均線、RSI、William、BIAS等技術指標，也使用不同天日的收盤價與移動平均線的差值、取絕對值、不取絕對值等方式探求最能表現出該神經元特色的數值向量。

This thesis proposes an improved Self-Organizing Maps (SOM) with Grey Relational Grade (GRG) by using the novel algorithm clustered financial time series and Fuzzy Rules to correct potential errors. The model is applied to several real markets to examine accuracy of the stock price prediction.
The characteristic of Grey System Theory is to extract useful information from either known or unknown data. The basic concept of Grey Relational Analysis is to judge the relational degree of data series by grouping similar degree of its geometry shape. This thesis replaces Minimum Euclidean Distance or Maximum Inner Product by proposing a winner method using Grey Relational Grade. The process of finding the largest GRG which is also the winner neuron is feedbacked for further process until a steady state is reached.
Each neuron contains a data series from several technical indicators, including Moving Average Line, RSI, William, and BIAS. A new technique indicator is also used based on several kinds of different attributes in closing price and moving average, and absolute values to present the neuron’s characteristics.

[1]The Economist， http://www.economist.com/opinion/displaystory.cfm?story_id=13022067，(2009/1/29).
[2]苦勞網，http://www.coolloud.org.tw/node/7663，(2007)。
[3] Robert L. Harvey, Neural Network Principles, pp. 1 (1994).
[4]黃綺年，「統計方法與類神經網路應用於國內開放式股票型基金投資績效分類及投資報酬率預測之研究」，國立成功大學統計學研究所(2004)
[5] Mark O. Afolabi, Olatoyosi Olude, “Predicting Stock Prices Using a Hybrid Kohonen Self Organizing Map (SOM)”, Proceedings of the 40th Hawaii International Conference on System Sciences (2007)
[6] Carlos Serrano-Cinca, “Self organizing neural networks for financial diagnosis”, Decision Support Systems, 17: pp.227–238 (1996).
[7]何鴻聖，「自我組織神經網路在選股策略的應用」，國立東華大學國際經濟研究所(2004) 。
[8] Marijana Zekic, “Neural network applications in stock market predictions: A methodology analysis”, In B. Aurar and R. Varazdin Logozar, editors, Proceedings of the 9th International Conference on Information and Intelligent Systems, pp 255–263 (1998)
[9] Y. Yoon, T. Guimaraes, and G. Swales, “Integrating artificial neural networks with rule-based expert systems”, Decision Support Systems, 11:pp.497–507 (1994)
[10] Chueh-Yung Tsao, “Financial Modeling Based on the Trajectory Domain”, National Central University, Taiwan (2004)
[11] Sergio Valero Verd, Mario Ortiz Garca, Carolina Senabre, Antonio Gabaldn Marn, Member, IEEE, and Francisco J. Garca Franco, “Classification, Filtering, and Identification of Electrical Customer Load Patterns Through the Use of Self-Organizing Maps”, IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 21, NO. 4 (2006)
[12] Chueh-Yung Tsao, Shu-Heng Chen, “Self-Organizing Maps as a Foundation for Charting or Geometric Pattern Recognition in Financial Time Series”, CIFEr’03 HONG KONG (2003)
[13]棗厥庸，「如何以金融市場的交易軌跡建立財務計量模型」，博士論文，國立中央大學財務金融研究所，桃園(2004)。
[14]Dong-Jun Yu,Yong Qi,Yong-Hong Xu,and Jing-Yu Yang,
"Kernel-SOM Based Visualization of Financial Time Series Forecasting", Proceedings of the First International Conference on Innovative Computing, Information and Control (ICICIC'06) (2006)
[15]Adrian Costea, Iulian Nastac, “Assessing The Predictive Performance Of Artifical Neural Network-Based Classifiers Based On Different Data Preprocessing Methods, Distributions And Training Mechanisms”, INTELLIGENT SYSATSESMESS SIINN GA CPCROEUDNICTTIINVGE, PFEINRAFNOCREM AANNCDE M OAFN AANGNE-MBEANSTED CLASSIFIERS 217 Intell. Sys. Acc. Fin. Mgmt. 13, pp.217–250 (2005)
[16] Y.-F.Wang, “Predicting stock price using fuzzy grey prediction system”, Chang Gung Institute of Nursing, Taiwan (2002).
[17]Cheng-Lung Huang,ChengYi Tsai, “A hybrid SOFM-SVR with a filter-based feature selection for stock market forecasting”, Expert Systems with Applications (2008)
[18]Chung-Chian Hsu,” Generalizing Self-Organizing Map
for Categorical Data”, IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 17, NO. 2 (2006)
[19]劉思峰、黨耀國、方志耕等，灰色系統理論以及其應用，第三版，科學出版社，北京 (2004)。
[20] http://www.researchinformation.co.uk/grey.php，(2009)。
[21]王志均，笑笑看線圖，早安財經文化，台北，第 34-49頁 (2006)。
[22] John J. Murphy, Technical Analysis of the Ftures Markets – A Comprehensive Guide to Trading Methods and Applications, New York Institute of Finanance (1986).
[23]http://stockcharts.com/school/doku.php?id=chart_school:technical_indicators，(2009)。
[24] L.A. Zadeh, “Fuzzy Sets”, Information and Control 8, pp.338 -353 (1965).
[25]石其生，「期貨與選擇權程式交易平台之實現」，碩士論文，台灣科技大學，台北(2008)。
[26 ] AiSM，臺灣科技大學資訊工程所人工智慧金融交易研究室， http://aism.csie.ntust.edu.tw，(2009)。
[27]張兆鴻，「Design of A Taiex Trade」，碩士論文，台灣科技大學，台北(2006)。
[28]林俊吉，「應用灰聚類進行台股指數期貨程式交易設計」，碩士論文，台灣科技大學，台北(2005)。
[29] Bao Rong Chang, Hsiu Fen Tsai, “Forecast approach using network adaptation to support vector regression grey model and generalized auto-regressive conditional heteroscedasticity”, Expert systems with Applications 34, pp.925-934 (2008)。
[30]田自力、陳碩珮，「GM(1,1)模型之傅立葉級數殘差修正法」(1996)。