臺灣博碩士論文加值系統

中文摘要
在統計學上，一般的統計分析，多偏重於兩變數間的相關性，而無法表示其變數間之因果關係，以身高體重為例，當身高愈重者，他的體重或許也愈重，兩變項間呈現正相關，然而在事實上，是身高影響體重？抑體重影響身高？若改變一個人的體重並不會影響其身高，所以體重並非是影響身高的原因。但在現實生活或社會科學的研究上，有許許多多的變數，其因果關係並非如上例那麼顯而易見，故在變項上，何者為因？何者為果？一直是學者很想探究的一個議題。
雖然貝氏網路近年來在各種研究上，藉由專家知識結構及學習模式，在因果關係及診斷上扮演著重要角色，亦被廣泛使用於醫學、資料探勘等領域之研究。
然而因果關係的推斷大都來自理論及長時間之觀察或實驗之資料，且貝氏網路目前並不能控制潛在變項，須藉由專家知識來建立，這對社會科學的研究似乎緩不濟急，故從資料中發現可能存在的潛在變項及因果關係便相對的重要。
　　本論文將在d隔離及淨相關的條件下，分析各個觀察變項並分群至同一潛在變項，以建立符合結構方程模式之測量模式，並結合PC2演算法及條件機率，仿建置貝氏網路之方法，建構以貝氏網路為基礎之結構方程模式。

Abstract
In statistics, general statistical analysis stresses on the relevance between the variables. But it is unable to express the causality between the variables. Take the height and body weight as example. Height higher, his body weight is also heavier. It is connected between the variables. As a matter of fact, however, does the height affect the body weight or the body weight affects the height? If the change of the body weight would not affect the height, it is not the cause that the weight can be affected by the height. There are lots of the variables in real life or in the researches of social science. The causality is not as obvious as the example that we have mentioned. On variable, which one is a cause? Which one is a result? Therefore, it is always an issue that the researchers like to find out in the statistic.
By means of the constructions of specialists and the mode of learning, the Bayesian networks play the important roles in the causality and the diagnoses, and they are used generally in the researches of medicine and the data survey, too. However, the educts of the causality come from the theories, the observations of a long time and the data of the laboratory. In the meantime, the Bayesian networks cannot control the latent variables so far. It must be constructed by specialists’ knowledge. These results in the research are too late to the social sciences. Therefore, it is very important to find out the latent variables and the causality that could be existed possibly.
In order to build a measurement model that matches the structural equation model, these variables will be analyzed individual and observed then to be divide into the same latent variables under the d-separation and partial correlation condition. By combining PC2 algorithm and conditional probability and after the Bayesian Networks, the structural equation model will be built base on Bayesian Network.

目 錄
第一章 緒論 1
1.1 研究動機 1
1.2 研究目的 3
1.3 論文架構 3
第二章 文獻探討 4
2.1 因果關係 4
2.2.1 貝氏網路的優點與性質 8
2.2.2 d 隔離(d-separation) & V 結構(V-Structure) 8
2.2.3馬可夫因果假設(Causal Markov Assumption) 11
2.3 貝氏網路建置的方法 12
2.3.1 CB演算法(CB Algorithm) 13
2.3.2 Ｋ2演算法(Ｋ2 Algorithm) 15
2.3.3 PC2演算法(PC2 Algorithm) 16
2.3.4 Scoring Function 18
2.4 科技接受模式 20
2.5 結構方程模式(SEM) 22
2.5.1測量模式 22
2.5.2結構模式 23
2.5.3因果性質 23
第三章 研究方法與設計 24
3.1 研究方法 24
3.1.1 ILV演算法(Identifying Latent Variables Algorithm) 24
3.1.2 網路建置的方法 26
3.2 研究流程 27
3.3 研究工具 28
3.4 研究對象與限制 28
第四章 研究結果 29
4.1.1 ILV演算法建立之測量模式 29
4.1.2 PC2演算法建立之結構模式 31
4.2 模擬實驗二 37
4.2.1 ILV演算法建立之測量模式 37
4.2.2 PC2演算法建立之結構模式 38
第五章 結論與建議 45
參考文獻 46
誌 謝 48

參考文獻
[1] 林育仕(2005)。利用基因網路分析變數間之因果關係，碩士論文，國立交通大學統計學研究所。
[2] 吳明隆(2006)。SPSS統計應用學習實務：問卷分析與應用統計(三版)。
台北：知城。
[3] 洪新原、梁定澎、張嘉銘(2005)。科技接受模式之彙總研究。資訊管理學報，
2(4)，211-234。
[4] 邱皓政(2005)。結構方程模式。台北：雙葉。
[5] 黃芳銘(2004)。社會科學統計方法-結構方程模式。台北：五南。
[6] 黃意策(2002)。應用資料採礦技術於貝氏網路架構之研究—線性代數與圖論之結合，碩士論文，東海大學工業工程與經營資訊研究所。
[7] 潘俊帆(2002)。以貝氏網路為基礎的個人化差異知識處理機制，碩士論文，東海大學工業工程研究所。
[8] 鄭祐昇(2004)。運用資料探勘技術於知識圖之建立，碩士論文，東海大學工業工程與經營資訊研究所。
[9] Bollen, K.A. (1989). Structural Equations with Latent Variables. John Wiley and Sons, New York, NY.
[10] Borooah, V. (2002). Logit and Probit: Ordered and Multinomial Models, Sage Publications, Thousand Oaks, CA.
[11] C. Glymour, R. Scheines, P. Spirtes,and K. Kelly.( 1987) Discovering Causal Structure: Artificial Intelligence,Philosophy of Science and Statistical Modeling.Academic Press.
[12] Cooper, G. and E. Herskovits . (1992) . A Bayesian Method for the Induction of Probabilistic Networks from Data. Machine Learning, 9, 309-347.
[13] Davis, F.D. (1989). Perceived Usefulness, Perceived Ease of Use and User Acceptance of Information Technology. MIS Quarterly, 13, 3, 319-340.
[14] Druzdzel, M. and H. Simon, (1993). Causality in Bayesian Belief Networks. In the Proceedings of the 9th Annual Conference on Uncertainty in Artificial Intelligence, 3-11.
[15] Friedman, N., M. Linial, I. Nachman, D. Peer, (2000) . Using Bayesian networks to analyze expression data. Proceedings of the International Conference on Computational Molecular Biology, 127-135
[16] Heckerman D. ( 1996). A tutorial on learning Bayesian networks. Technical Report MSR-TR-95-06, Microsoft Research, Redmond, Washington. Online: http://citeseer.ist.psu.edu/heckerman95tutorial.html.
[17] Lee, B., A. Barua, and A.B. Whinston. (1997). Discovery and representation of causal relationships in MIS Research: A methodological framework. MIS Quarterly, 21, 1, 109-136.
[18] Mulaik,S.A.,& Jame,L.R.(1995). Objectivity and reasoning in science and structural equation modeling. In:R.H.Hoyle(Ed.),Structural equation modeling :Concepts, issues and application(pp.118-137).Thousand Oaks,CA:Sage.
[19] Pavlou, P.A.(2003). Consumer Acceptance of Electronic Commerce: Integrating
Trust and Risk with the Technology Acceptance Model. International Journal of
Electronic Commerce, 7, 3, 69-103.
[20] Pearl, J. and T. Verma, (1991). A Theory of Inferred Causation. Proceedings of Princliples of Knowledge Presentation and Reasoning, 441-452.
[21] Pearl, J., (1995).Causal Diagrams for Empirical Research. Biometrika, 82, 4, 669-710.
[22] Spirtes, P., C. Glymour, C. and R. Scheines.(2002). Data mining tasks and methods: Probabilistic and casual networks: mining for probabilistic networks, Handbook of data mining and knowledge discovery, Oxford University Press, New York, NY.
[23] Scheines, R., P. Spirtes, C. Glymour, C. Meek, and T. Richardson. (1998). TETRAD Project: Constraint Based Aids to Causal Model Specification. Multivariate Behavioral Research, 33, 1, 119-128.
[24] Silva, R. Scheines, C. Glymour, and P. Spirtes. (2003). Learning measurement models for unobserved variables. Proceedings of 19th Conference on Uncertainty in Artificial Intelligence, pages 543–550.
[25] Spirtes, P., T. Richardson, C. Meek, R. Scheines, and C. Glymour, (1998). Using path diagrams as a structural equation modeling tool. Sociological Methods and Research, 27, 2 , 182-225.
[26] Spohn, W.(2000). Bayesian Nets Are All There IS To Causal Dependence. In M.C. Galavotti (Ed.), Stochastic Dependence and Causality, CSLI Publications, Stanford, CA.
[27] Verma, T. and J. Pearl .(2003). An algorithm for deciding if a set of observed
independencies has a causal explanation, Proceedings of 8th Conference on
Uncertainty in Artificial Intelligence, 323-330.
[28] Zheng E. & P.A. Pavlou. (2005).Structural Equation Modeling with Causal Relationships: A Bayesian Networks Methods for Inferring Causality from Data. under review in Management Science.