本研究的目的在於從民航飛安檢查員的日常飛安查核結果中，發掘潛在關聯規則。查核資料以月份為單位，將安全相關狀態彙總整理計算出查核不滿意率。在準備分析資料中，先清除多餘不需要的資料，並應用修正後的MVC(Missing Values Completion)法來處理屬性資料的遺漏問題。而修正後的MVC法使用SOM(Self-Organizing Map)群聚技術來將資料進行分群。在同ㄧ群的資料紀錄中擁有著相似的資料型態。根據假設，以同ㄧ群集中計算出的beta平均值來填補遺漏項目。 我們使用Agrawal et al. (1993)提出之Apriori 關聯規則演算法來分析資料。由於Apriori演算法無法處理數值資料，因此在使用該演算法之前，將績效指標根據統計處理控制技術轉換成為正常與非正常之邏輯形態。除此之外，亦使用傳統的Pearson Correlation Analysis來了解飛安事件與飛安檢查結果之關聯。在本研究中，將考慮「時間遞移」的問題，並從中找出之間的關聯性。

The purpose of this research is to discover any potential association rules for aviation safety inspection results which are performed daily by CAA aviation safety inspector. The inspection data will be aggregated to identify the unfavorable rate for each safety related performance in one month period. To prepare the analyzed data, we clean the redundant data and apply a modified MVC（Missing Values Completion）method to deal with attribute value missing. The modified MVC method uses the SOM (self-organization map) clustering technology to classify data records into clusters. The data records in the same cluster have similar data pattern. According to the assumption, the beta mean value in the same cluster is calculated to fill into the missing attribute. We applied the Apriori association rule algorithm described by Agrawal et al. (1993) to the analyzed data. Since the Apriori algorithm does not process numerical data, we transform the performance attributes to the set of discrete categories, normal and abnormal, by a statistic process control technique before application of the algorithm. Besides, the traditional Pearson correlation analysis has also been conducted to figure out the relationship between aviation events and safety inspection results. In our research, time lag has been considered as an important issue to discover such a relationship.

目錄

第一章	緒論	1
1.1	研究背景	1
1.2	研究目的	3
1.3	研究流程與架構	4
1.4	論文架構	7
第二章	文獻探討	8
2.1	飛安相關研究	8
2.2	資料探勘方法	11
2.3	分群技術	13
2.4	關聯分析	26
2.5	遺漏值的處理	33
第三章	模式建構	36
3.1	資料建構	36
3.2	查核指標關聯模式	44
3.3	飛安事件與績效指標關聯性分析	47
第四章	飛安績效指標關聯性分析	50
4.1	實作資料來源與工具	50
4.2	遺漏值填補結果	50
4.3	相關性分析	52
4.4	APRIORI關聯分析	57
第五章	結論與未來研究	67
參考文獻	69

表目錄
表2- 1  飛航安全相關文獻彙整	10
表2- 2  分群技術一覽表	13
表2- 3  分群技術比較表	15
表2- 4  SOM相關研究主題	24
表2- 5  關聯規則應用表	31
表2- 6  遺漏值填補法之比較表	34
表3- 1　查核績效指標一覽表	37
表3- 2  輸入資料與神經元對照表	43
表3- 3  神經元統計歸納表	43
表4- 1  全部查核項目相關項目	53
表4- 2  適航相關項目	53
表4- 3  航務客艙相關項目	54
表4- 4　不考慮時間遞移之飛安事件主題與績效指標之相關項目	54
表4- 5　不考慮時間遞移之飛安事件種類與績效指標之相關項目	54
表4- 6  考慮時間遞移之飛安事件主題與績效指標之相關項目	55
表4- 7　考慮時間遞移之飛安事件種類與績效指標之相關項目	56
表4- 8  適航查核項目關聯規則	58
表4- 9  航務客艙查核項目關聯項目	59
表4- 10全部事件APRIORI關聯規則分析結果	60
表4- 11重要事件APRIORI關聯規則分析結果	61
表4- 12 航管事件APRIORI關聯規則分析結果	62
表4- 13 航務事件APRIORI關聯規則分析結果	62
表4- 14 場站事件APRIORI關聯規則分析結果	63
表4- 15 地面事件APRIORI關聯規則分析結果	63
表4- 16其他事件APRIORI關聯規則分析結果	64
表4- 17當期與未來影響力之飛安績效指標統整表	66

圖目錄 
圖1- 1  國籍航空與IATA 全球失事統計比較	1
圖1- 2  研究架構圖	6
圖1- 3  論文架構	7
圖2- 1  乳酪理論代表圖	9
圖2- 2　失事事件形成因素與系統安全概念	9
圖2- 3  生物神經元	16
圖2- 4  類神經基礎架構	16
圖2- 5  SOM架構圖	18
圖2- 6  高斯型式之鄰近區域函數	19
圖2- 7  映射圖	19
圖2- 8　常見鄰近區域類型	22
圖2- 9　模擬交易紀錄	27
圖2- 10  APRIORI演算法	29
圖3- 1   MVC流程圖	41
圖3- 2　 修正後之遺漏值填補流程	41
圖3- 3　 檢查項目關聯分析架構	47
圖3- 4　 飛安事件與檢查項目關聯分析架構圖	49
圖4- 1   多種分析參數比較圖	52
圖4- 2　  比較圖	52
圖4- 3　不考慮時間遞移之飛安事件與查核項目之關聯規則累計次數圖	65
圖4- 4　考慮時間遞移之飛安事件與查核項目之關聯規則累計次數圖	65

參考文獻 
中文文獻 
[1] 交通部民用航空局,民航政策白皮書 ,2000 
[2] 交通部民用航空局,航務檢查手冊v3.0 ,2000 
[3] 交通部民用航空局,適航檢查手冊v3.0 ,2000 
[4] 交通部運輸研究所,「台灣地區飛航安全概述」,1996 
[5] 交通部運輸研究所,「國內外航空事故肇因分析與失事調查組織以及作業之研究」,1997 
[6] 汪進財、葉文健,「航空公司飛安管理與分析系統之建立」,國科會計畫(NSC89-2211-E-009-080)部分成果,2001。 
[7] 張有恆、李昭蒂,「航空公司航安全績效評估之研究」,2001,p.6-1 
[8] 喬志弘,「航空安全人為因素」,飛行安全季刊,5,第九期,1999,p.50 
[9] 曾憲雄等人,「資料探勘」,2005,旗標 
[10] 葉怡成,「類神經網路模式應用與實作」,2000,儒林 
[11] 鐘平祥,「我國民航業者與主管機關提昇飛航安全策略之研究」, 國立交通大學，碩士論文，2001 

英文文獻 
[1] Agrawal, R. and Srikant, R., “Fast Algorithm for Mining Association Rules” , Proc. Conf. Very Large Databases, Sept, 1994, p. 487-499. 
[2] Allinson, N., Yin, H., & Obermayer, K. , “Introduction: new developements in self-organizing maps” ,Neural Networks, 15, 2002, p. 943. 
[3] Barnett, A. and Higgins, M. , “Airline Safety: The Last Decade”, Management Science, vol.35, 1990, p.1-21. [4] Besselink, C., “Integrated Safety Management System” , Presented at the Flight Safety Foundation’s 49th Annual International Air Safety Seminar at Dubai, 1996. 
[5] Boeing, “1999 Statistical Summary of Commercial Jet Airplane Accidents”, 2000. 
[6] Brin, S. and Silverstein, C., “Beyond Market Baskets: Generalizing Association Rules to Correlations” , ACM SIGMOD Conference on Management of Data, 1997, p.265-276. [7] Cohen, E., Datar, M.and Fujiwara, S., “Finding Interesting Associations without Support Pruning”, IEEE Trans. On Knowledge and Data Engineering, Vol. 13, No. 1, 2001, p. 64-78. 
[8] Dempster, A.P., Laird, N.M., Rubin, D.B., “Maximum likelihood from incomplete data via the EM algorithm”, Journal ofthe Royal tatistical Society, Series B Methodological 39 (1), 1977, p.1–38. 
[9] Diaz, R.I.and Cabrera, D., “Safety Climate and Attitude as Evaluation Measures of Organizational Safety,” Accident Analysis and Prevention,29,5, 1997 ,p.643-650. 
[10] Dionne, G., Gragne, R., Gagnon, F., Vanasse, C.and Debt, “Moral Hazard and Airline Safety: An Empirical Evidence,” Journal of Economics,79, 1997, p.397-402. 
[11] Dittenbach, M., Merkl, D. & Rauber, A., “The growing hierarchical self-organizing map”, In S. Amari, C.L. Giles, M. Gori & V. Puri, (Eds.), Proceedings of the Int. Joint Conference on Neural Networks (IJCNN 2000), vol. VI, 2000,pp.15-19). Como, Italy: IEEE Computer Society. 
[12] Edkins, G.D., ”The INDICATE safety program: evaluation of a method to proactively improve airline safety performance ,” Safety Science, 30, December ,1998 , p249-308.
[13] Evans,W.N., “Deregulation and Airline Safety: Evidence from Count Data Models mimeo ,” University of Maerland-College Park, 1989. 
[14] Frawley, W.J. , “Knowledge Discovery in Data Base : An Overview,” Menlo Park ,CA : AAAI/Press , The MIT Press, 1991. 
[15] Golbe, D.L., “Safety and Profits in the Airline industry ,” Journal of Industry Economics, 34, 1989, p.305-318. 
[16] Guerrero-Bote, V. P., López-Pujalte, C., Moya-Anegón, F. D., & Herrero-Solana, V. , “Comparison of neural models for document clustering.”, Int. Journal of Approximate Reasoning, 34, , 2003, p.287-305. 
[17] Gustavo E. A. P. A. Batista & Monard, M.C., “An Analysis of Four Missing Data Treatment Methods for Supervised Learning ,”Applied Artificial Intelligence, vol.17, no.5-6, 2003, p. 519-533. 
[18] Hasson, J., ”Boeing’s Safety Assessment Processes for Commercial Airplane Designs” IEEE, 1997, p.4.4-1 -4.4-7. 
[19] Haykin, S., Neural Networks, 1994, Macmillan 
[20] Heinrich, H.W., “Industrial Accident Prevention. ” McGraw-Hill, New York, 1950 ,p. 10-32. 
[21] Jain, A. K., Murty, M. N., & Flynn P. J. (1999). Data clustering: a review. ACM Computing Surveys, 31(3), 264-323. 
[22] Jiawei, H. and Micheline, K. ,Data Mining:Concepts and Techniques , 2001, John Wiley & Son. 
[23] Kanafani, A. and Keeler, T.E., “Air Deregulation and Safety: Some Econometric Evidence From Time Series,” Logistics and Transportation Review, 26, 3, 1990, p.203-209 
[24] Ke, W., Yu, H. , & Jiawei, H., “Mining Frequent Itemsets Using Support Constraints,”Proc. Int. Conf. on on Very Large Data Bases, 2000 ,p.43-52. 
[25] Kiang, M. Y. , “Extending the Kohonen self-organizing map networks for clustering analysis.”, Computational Statistics & Data Analysis, 38,2001, p.161-180. 
[26] Koh, J., Suk, M., & Bhandarkar, S. M., “ A multiplayer self-organizing feature map for range image segmentation”, Neural Networks, 8(1), 1995,67-86. 
[27] Kohonen, T., “The Self-Organizing Map,” Proceedings of the IEEE, vol. 78, no.9, 1990, p. 1464-1480.
[28] Kohonen, T., Kaski, S.& Lagus, K. , “Self organization of a massive document collection” , IEEE Transactions on Neural Networks, Special Issue on Neural Networks for Data Mining and Knowledge Discovery, 11(3), 2000, p.574-585. 
[29] Kwong, W. et al., “Mining association rules with weighted items,” Database Engineering and Applications Symposium, 1998. Proceedings. IDEAS'98. International, p.68 –77. 
[30] Liu, B., Hsu, W. and Ma, Y., “Mining Association Rules with Multiple Minimum Supports”, SIGKDD International Conference on Knowledge Discovery & Data Mining(KDD-99), August 15-18, 1999, San Diego. 
[31] Matthews, S., “Human Factors in Aviation Safety,” 10th-nationality aviation safety annual conference,2000, Taipei 
[32] Mao, J. & Jain, A. K., “A self-organizing network for hyperellipsoidal clustering (HEC)”, IEEE Transactions Neural Networks, 7,1996,p.16-29. 
[33] Mangiameli, P., Chen, S. K., & West, D. , “A comparison of SOM neural network and hierarchical clustering methods.”, European Journal of Operation Research, 93, 1996, p.402-417. 
[34] Ng, R. & Han, J. , “Very large data bases” , Proceedings of the 20th Int. Conference on Very Large Data Bases (VLDB’94, Santiago, Chile), VLDB Endowment, Berkeley, CA, September,1994, p.144-155. 
[35] Pak, C. W., Whitney ,P. & Thomas,J. ,“Visualizing association rules for text mining, Information Visualization”,Proceedings of IEEE Symposium on Info Vis '99, 1999, p.120 -123, p152. 
[36] Park, J. S., Chen, M. S., and Yu, P. S., “An Effective Hash Based Algorithm for Mining Association Rules”, Proc. of ACM SIGMOD, May 23-25, 1995, p175-186. [37] Pooley, E.D., “Putting air safety management into practice demands a positive corporate safety culture” , ICAO Journal, vol.54,no.1, 1999,p.10-14. 
[38] Pyle, D.,“Data Preparation for Data Mining”, 1999, Morgan Kaufmann Publishers. 
[39] Ragel, A., Cremilleux, B., “MVC-a Preprocessing Method to Deal with Missing Values,” Knowledge-Based Systems, 1999, p.285-291. 
[40] Rauber, A., Paralic, J., & Pampalk, E.,Empirical evaluation of clustering algorithms, 2000, Vienna University of Technology, Department of Software Technology.
[41] Reason, J., Managing the Risks of Organizational Accidents, 1997,Ashgate Published, USA. 
[42] Ressom, H., Wang, D., & Natarajan, P. , “Adaptive double self-organizing maps for clustering gene expression profiles”, Neural Networks, 16, 2003, p. 633-640. 
[43] Sa,u D. L.and David, W. C., “Maintenance of Discovered Association Rules: When to update,” Proc. of SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery, May, 1997. 
[44] Saaty, T.L., The Analytic Hierarchy Process, 1980,McGraw-Hill, New York. 
[45] Sangole, A. & Knopf, G. K., “Visualization of randomly ordered numeric data sets using spherical self-organizing feature maps”, Computers & Graphics, 27, 2003, p.963-976. 
[46] Sau, D. L., David, W.C. ,“Maintenance of Discovered Association Rules: When to update? ”, Proc. of SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery, May, 1997. 
[47] Sharma, S.C. ,Applied multivariate techniques,1996, John Wiley & Sons. 
[48] Shin, M., TsenGand, K. H. , “A Pre-processing method to deal with missing values by integrating clustering and regression techniques”, Applied Artificial Intelligence, 2003, 17:535–544. 
[49] Shouhong, W. , “Application of self-organizing maps for data mining with incomplete data sets”, Neural Comput & Applic, 2003, 12:42–48. 
[50] Shouhong, W. , Hai, W., “Conceptual construction on incomplete survey data”, Data &Knowledge Engineering, 49, 2004, p311-323. 
[51] Thom, T., “Human Factors and Pilot Performance-Safety”, 1997, First Aid and Survival, Airlife Published, England. 
[52] Vesanto, J., Himberg, J., Alhoniemi, E., & Parhankangas, J. , “SOM toolbox for Matlab 5 “, Espoo, Finland: Helsinki University of Technology, April, 2000 , SOM Toolbox Team. 
[53] Wur, S. Y. , Leu, Y., “An Effective Boolean Algorithm for Mining Association Rules in Large Database” , DASFAA, 1999, p.179-186. 
[54] Yin, H. , “Data visualization and manifold mapping using the ViSOM”, Neural Networks, 15, 2002, p. 1005-1016. 
[55] Zhang, X., Li, Y., “Self-Organizing Map as a New Method for Clustering and Data Analysis.”, International Joint Conference on Neural Networks, 1993, p. 2448-2451.
[56] Zhang, T., Ramakrishnan, R., & Livny, M., “BIRCH: An efficient data clustering method for very large databases. ”, In Proceedings of the ACM SIGMOD Conference on Management of Data, Montreal, Canada, 25(2) , June,1996,p. 103-114.