臺灣博碩士論文加值系統

群集分析(Clustering Analysis)為資料探勘(Data Mining)技術中所廣泛被人們所應用的技術之一，目前已被大量應用在商業顧客分析、醫學、病理學、自然科學等領域，近年來更大量用在基因資料分析或影像分析等領域。目前人們使用群集分析所探討的資料種類絕大部分為類別性資料(Qualitative data)、數量型資料(Quantitative data)或兩種資料混和型資料。較少有人討論到是否可以用群集分析的理論來探討流程型資料(Process Data or Flow Data)。本研究旨在建立能夠分析具有流程型資料之物件其相似度模型，發現將模型建立在群集分析之匯聚型階層式群集分析法(Agglomerative Hierarchical Method)之上，並使用Average Linkage計算兩物件之距離，可得到較佳之分群(Grouping)效果。但流程順序之相似度(similarity)判定需視不同研究標的給予不同的相似度得分，使用者須端看研究環境選擇較適合之判定準則。
本研究所挑選之個案資料為某發動機修護工廠執行GE CF6-80C2 Engine 翻修(Overhaul)工作時，維修過程中所產生之自修工件，希望能得到自修工件之間彼此在維修流程上之相似程度，以利管理階層於做工作站安排工作時或是其他管理決策之依據，例如修理站人員專長規劃、新維修能量開發標的甚至是未來維修站新購機台之考量。
最後得到的結果發現此分群模型可將發動機翻修時所產生的205項自修工件在經過流程相似度分群之後可得到8個較明顯之工件群組。在經過與現場工作人員討論後，可給予各個工件群組依維修流程上之相似意義，可見集群分析概念亦可替流程性類型資料做相似度分群，達到物件分群效果。

Cluster analysis is one of the data mining methods that are commonly applied in the fields including business analysis, medicine, pathology, and natural science. In recent years, it has also been largely applied to the genetic data analysis or image analysis fields.
Nowadays, the majority of people who apply the cluster analysis technique are on qualitative data, quantitative data or mixed data types and it’s noted that fewer people discuss whether the cluster analysis theory can also be applied on process data or flow data.
The aim of this thesis is to establish a similarity model that contains the objects of process data when applying cluster analysis. It’s found that a better grouping effect can be achieved when the model is based on the agglomerative hierarchical method of the cluster analysis and that the average linkage is used when calculating distance between objects.
Nevertheless, the assessment of similarity of process order requires different scores depending on individual study subject; therefore users need to set appropriate criteria that cater for their specific research environments.
The case study used in this thesis is based on the data collected from an engine maintenance plant during the overhaul process on the GE CF6-80C2 Engine. Several in-house repair parts are being identified in this process and by studying the of similarity degree of these in-house repair parts, it’s hoped that the findings can be useful to the management in planning shifts or making other managerial decisions including competency planning for maintenance technicians, determining new maintenance capacity or acquisition of maintenance machinery.
In conclusion, the study shows that after applying this similarity model, there are 205 in-house repairing parts identified in the engine overhaul process and those can be grouped into 8 major clusters. Having discussed further with production staff, each group can be given a similar meaning that corresponds to its overhaul process. As such, it shows that cluster analysis theory can also be applied in the process data type when carrying out grouping analysis based on their similarities.

第一章 緒論.................................................1
1.1 研究環境................................................1
1.2 研究目的................................................1
1.3 研究動機................................................2
第二章 相關文獻與產業探討.....................................3
2.1 資料探勘................................................3
2.2 群集分析................................................3
2.3 航空發動機修護流程.......................................4
2.4 發動機零附件修護流程.....................................5
第三章 模型建立.............................................7
3.1 資料蒐集................................................8
3.2 環境設定................................................8
3.3 資料整理...............................................10
3.4 建立相似度規則.........................................12
3.4.1 依兩物件中相同程序站數量給分...........................12
3.4.2 依兩物件中連續相同程序站加權給分........................12
3.4.3 依程序站位置相同加權給分...............................12
3.4.4 其他規則可視研究環境訂定...............................13
3.5 製作相似度矩陣.........................................13
3.6 資料分群...............................................14
3.7 結果分析...............................................16
第四章 個案分析............................................19
4.1 前言..................................................19
4.2 資料整理...............................................19
4.2.1 工作站設定...........................................19
4.2.2 資料蒐集及整理.......................................20
4.2.2.1 資料蒐集...........................................20
4.2.2.2 去除每張工作單所共通站別.............................22
4.2.2.3 合併工作單程序中連續相同站別.........................22
4.2.3 工作單經整理後表示方法................................24
4.3 工作單合併與分類.......................................24
4.3.1 完全相同工作順序之單合併...............................24
4.3.2 工件依總作站數量分類..................................25
4.4 設立相似度得分規則 設立相似度得分規則.....................26
4.4.1 兩工件中單一站相同得 1分..............................26
4.4.2 兩工件中連續2工站相同得2分，以下連續 3、4、5站相同類推....27
4.4.3 總得分..............................................27
4.5 製作相似度矩陣.........................................28
4.5.1 1-4 站短工件相似度矩陣................................28
4.5.2 5-10 站長工件相似度矩陣...............................28
4.6 資料分群...............................................28
4.6.1 1-4 站短工件分群.....................................28
4.6.2 5-10 站長工站分群....................................32
4.7 結果分析...............................................44
4.7.1 1-4站短工站工件件分群結果.............................44
4.7.1.1 1-4站短工站工件件群組 1.............................44
4.7.1.2 1-4站短工站工件件群組 2.............................46
4.7.1.3 1-4站短工站工件件群組 3.............................48
4.7.1.4 1-4站短工站工件件群組 4.............................49
4.7.1.5 1-4站短工站工件件群組分析...........................51
4.7.2 5-10 站長站工件工件分群結果...........................51
4.7.2.1 5-10 站長工站工件件群組1............................51
4.7.2.2 5-10 站長工站工件件群組2............................53
4.7.2.3 5-10 站長工站工件件群組3............................54
4.7.2.4 5-10 站長工站工件件群組4............................56
4.7.2.5 5-10 站長工站工件群組分析...........................57
第五章 結論與建議...........................................58
參考文獻...................................................60

1.Han, J. and Kamber. M. “Data Mining：Concepts and Techniques,” Morgan Kaufmann Publisher, 2001.
2.Fayyad, M. U.(1996). Datamining and knowledge discovery：making sense out of data. IEEE Expert, 11(10), 20-25.
3.Fayyad, M. U., Piatetsky-Shapiro, G. & Smyth, P.(1999). “From data mining to knowledge discovery: an overview,” in advances in knowledge discovery and data mining, Fayyad, M. U., Piatetsky-Shapiro, G. & Smyth, P. & Uthurusamy, R., eds., AAAI/MIT Press,pp. 1-36
4.Frawley, W. J., Piatetsky-Shapiro, G., and Matheus, C. J., “Knowledge discovery in databases: An overview.” AI Magazine (13:3), pp. 57-70, 1992.
5.Grupe, F. H., and Owrang, M. M., “Database Mining Discovering New Knowledge and Cooperative Advantage,” Information System Management (12:4), pp. 26-30, 1995.
6.Jiawei, H., and Micheline, K., “Data Mining：Concepts and Techniques,” Morgan Kaufmann, pp. 59-60, 2001.
7.Grupe, F. H., and Owrang, M. M., “Database Mining Discovering New Knowledge and Cooperative Advantage,” Information System Management (12:4), pp. 26-30, 1995.
8.莊澤生,利用資料探勘技術發掘議題網路，國立中山大學資訊管理研究所碩士論文，2002年