在競爭激烈的LCD 產業中，世界各大面板廠必須不停地建造次世代的工廠，才能符合市場需求及降低生產成本。因為建廠投入的資金非常大，必須有效的控制成本、增加效率，才能在競爭中生存下來。而建廠工作中最重要的工作項目之一是為整合電腦整合製造(Computer Integrated Manufacturing – CIM)系統之建置，此一系統關係著該廠未來幾年之生產能否降低成本、增加效率及符合市場之須需求量。故CIM系統之製造容量為關鍵因素，而其中CPU之處理能力(capacity) 則為必須謹慎考慮預測的項目之一，必須能準確預測才能符合未來幾年之生產製造及擴廠之容量需求。
本研究旨在探討如何利用迴歸分析來建立LCD產業中CIM系統內CPU的預測模式(model)，如此便可以在規劃下一個世代廠時，做為重要的容量預測依據。研究上利用統計方法的多元迴歸分析，針對訓練期的已知資料進行分析研究以建立CPU的預測模式(model)。建立預測模式時，先利用趨勢迴歸線依時間序列取得CIM系統內關鍵參數(Key Parameters)的未來預估值，再把這些預估值代入由訓練期所建立的迴歸模式，最後便可取得未來N個月的CPU預測值。在預測誤差值方面，當系統內關鍵參數的趨勢迴歸線取R平方最大時，可以得到較小的預測誤差。

In the heavy competing LCD industry, to meet the market requirements and reduce production costs, large-scale and global known companies have to continuously build new generation LCD factories whenever necessary. Since the capital put in is so enormous that those companies have to proficiently control their production costs in order to be ahead in the competitions. One of the major works in building a LCD factory is to set up the Computer Integrated Manufacturing, i.e. the CIM, system that will determine whether the factory can lower production cost, improve production performance, and meet the market requirements or not. The capacity of a CIM, especially of its CPU, has to be estimated very precisely in order for the factory to run properly during its lifespan without major upgrade.
This research aims to investigate how to use Regression Analysis to build a predict model of CPU capacity of a LCD factory CIM system. This model can provide a vital reference for companies in formulating a new generation factory. By using the multiple Regression Analysis and focusing on the training data for a specific period, the CPU capacity predicted model is first launched. Then the Key Parameters are identified and their prediction values are gained by the trend line of the Regression in a chronological order. Feeding the prediction Key Parameters’values into the Regression patterns launched during the training periods, the CPU capacity prediction values of the future months can be obtained. For the prediction deviation, it is concluded that when the trend line of the Regression of the Key Parameters uses the largest value of the R square, the predicted deviation can be smaller.

參考文獻
[1] 李惠妍(2003)，類神經網路與迴歸模式在台股指數期貨預測，國立成功大學高階管理碩士在職專班，碩士論文。
[2] 李文偉(2002)，應用多元迴歸建立營建工程工率評估模式之研究－以橋樑工程為例，國立中央大學土木工程學系，碩士論文。
[3] 李博志(2003)，應用計量經濟學，五南圖書出版公司。
[4] 羅華強(2005)，類神經網路：MATLAB的應用，高立圖書有限公司。
[5] 榮泰生(2004)，活用EXCEL精通行銷研究，五南圖書出版公司。
[6] 唐莎莉，商業軟體應用（二），國立中央大學教學網，經濟系教材http://webclass.ncu.edu.tw/~tang0/
[7] Microsoft網站，利用圖表顯示趨勢和預測銷售，http://office.microsoft.com/zh-tw/excel/HA010877851028.aspx
[8] 王文中(2004)，統計學與Excel資料分析之實習應用，博碩文化股份有限公司。
[9] 李曉隆(2002)，出租公寓之租金價格預測－複迴歸分析與類神經網路的比較，國立台灣科技大學企業管理研究所，碩士論文。
[10] 朱正修(2004)，台灣股市與國際股市連動性之研究，國立成功大學統計學系，碩士論文。
[11] 蘇木春、張孝德(2004)，機器學習：類神經經網路、模糊系統以及基因演法則，全華科技圖書股份有限公司。
[12] 周文賢(2002)，多變量統計分析SAS/STAT使用方法，智勝文化出版公司。
[13] 吳宗正(1992)，迴歸分析:理論與應用，復文書局。
[14] 陳順宇(2007)，迴歸分析，華泰書局。
[15] 陳欽雨、李得盛(1998)，房租預測模型之建構與分析，明新學報，20期頁15-25。
[16] 楊雅媛(2002)，迴歸分析與類神經網路預測能力之比較，國立政治大學統計學系，碩士論文。
[17] 傅粹馨(1996)，多元迴歸分析中之結構係數與逐步迴歸，教育資料與研究期刊，11民85.07頁25-35。
[18] 陳品文(1998)，以多元迴歸模式建構股票評價模型，國立東華大學應用數學研究所，碩士論文。
[19] 陳順宇(1997)，迴歸分析，華泰書局。
[20] 姚漢禱(2001)，多元逐步迴歸分析在體育研究上的應用，中華民國大專院校九十學年度體育統計研討會，國立體育學院，36-65頁。
[21] 寇龍華(1979)，如何用多元迴歸法作商情預測，財政經濟月刋，29:1 頁8-16。
[22] 何宜鍵(1996)，上市公司股價報酬率及波動性之預測-類神經網路及多元迴歸模型之配適性研究，國立中正大學企業管理學系，碩士論文。
[23] 盛嘉宏(2004)，類神經網路與統計迴歸模式在暴潮之研究，國立臺灣海洋大學海洋科學系，碩士論文。
[24] Hair, Joseph F. (2006). Multivariate Data Analysis, Upper Saddle River, N.J. :Pearson Prentice Hall.
[25] Nigel Griffiths, IBM Performance Monitoring Tools website http://www-941.ibm.com/collaboration/wiki/display/WikiPtype/nmon
[26] Stephen A. Delurgio (1999). Forecasting Principles and Applications, University of Missouri Cansas City, First Edition.
[27] David F. Groebner (2007). Introduction to Linear Regression and Correlation Analysis. Ch. 14 in Business Statistics: A Decision-Making Approach, 7e, 2007.
[28] Ashish K. Sen, M. S. Srivastava (1990). Regression Analysis: Theory, Methods, and Applications., New York :Springer-Verlag.
[29] LIU M. C.; WAY KUO ; SASTRI T. (1995). “An exploratory study of a neural network approach for reliability data analysis.” Journal of Quality and reliability engineering international, vol. 11, no 2, pp. 107-112
[30] U.A. Kumar (2005). Comparison of neural networks and regression analysis: a new insight, Expert Systems with Applications, pp. 424–430.
[31] Bill, C. H., L. W. Rick, et al. (1994). Predicting graduate student success: a comparison of neural networks and traditional techniques. Computers Operations Res. 21(3): 249-263.
[32] Abdelmonem A. Afifi, Virginia Clark, Susanne May (2004). Computer-aided Multivariate Analysis, 4th Edition, New York: Chapman & Hall/CRC
[33] Mendenhall, William and Sincich, Terry (2003). A Second Course in Statistics: Regression Analysis, Upper Saddle River, N. J.: Prentice Hall.
[34] Alvin C. Rencher. (2002). Methods of multivariate analysis, 2nd ed. New York :J. Wiley.
[35] Thomas, B. (1973). System independent tracing for prediction of system performance. Proceedings of the 1st symposium on Simulation of computer systems. Gaithersburg,
Maryland, United States, IEEE Press.
[36] Seungwoo, H. and W. H. Daniel (2005). The use of simulation for productivity estimation based on multiple regression analysis. Proceedings of the 37th conference
on Winter simulation. Orlando, Florida, Winter Simulation Conference.
[37] Jon, D. C., J. R. Thomas, et al. (1979). "Computer performance evaluation: an empirical approach." SIGMETRICS Perform. Eval. Rev. 8(1-2): 97-101.
[38] John, W. B. and R. W. David (1975). "A Straightforward Model for Computer Performance Prediction." ACM Comput. Surv. 7(2): 73-93.
[39] Craig A. Mertler, Rachel A. Vannatta. (2005). Advanced and multivariate statistical methods: practical application and interpretation. Glendale, CA : Pyrczak, c2005.
[40] Yixin, C., D. Guozhu, et al. (2002). Multi-dimensional regression analysis of time-series data streams. Proceedings of the 28th international conference on Very
Large Data Bases. Hong Kong, China, VLDB Endowment.