應用空間資料探勘於未來需求規劃之研究－以都會區捷運系統為例

A Study on Use Spatial Data Mining for Planning Future Demand-Case of Urban Mass Rapid Transit System

10.29767/ECS.201003.0004

黃書猛(Shu-Meng Huang)；張中權(Chung-Chen Chang)

資料探勘 ； 都會區捷運系統 ； 集群分析 ； 疊合分析 ； data mining ； Urban Mass Rapid Transit System ； clustering analysis ； verlay analysis

Electronic Commerce Studies

8卷1期（2010 / 03 / 31）

105 - 121

繁體中文

資料探勘可藉由以往歷史資料之探索取得未來有用的資料，運用於未來需求之規劃，而都會區捷運系統之需求亦係基於大眾對公共運輸（空間）需求而衍生；因此，本研究從需求面探討相關歷史資料及大眾運輸特性確認資料探勘之範疇而取得既有公車系統、交通樞紐點及重要公共設施有其關聯性；其同質性之部份均為地理位置，故採用地理資訊系統(GIS)將該三項關聯要素之各個地點予以訂位疊合，惟因該三項要素功能性質不同，續借助訪談多位交通專家確立並取得其相互間關聯強度，並將該結果以層級分析法(AHP)分析該三項對站址選擇之權重引入GIS內，藉由群集分析得到適度之未來需求點，並尋得最符合捷運運輸需求之系統；為確認本研究方法之可行性，乃以人口逾150萬人之桃園都會區為例作驗證，該都會區捷運系統已依傳統之現場調查、居民問卷及所有交通系統分析等方式規劃完成，經依本研究之資料探勘程序施作後，所獲得之捷運系統與傳統方式之規劃結果頗為近似；故應具有可用性。

As data mining can get information for future use by exploring historical data and the need for Urban Mass Rapid Transit System is derived from the public demands for public transportation (space), this study correspondingly discusses the related historical data and the MRTS attributes from the demand aspect to confirm the data mining range and get the relevance in between the public transit system, the traffic junctions and important public facilities, which share a common attribute of geographic location. Therefore, we apply Geographic Information System (GIS) in overlaying these locations of the above three factors in relevance. However, as three factors are different in function, we shall confirm and get the relevance in between them by interviewing some traffic experts. And after conducting AHP analysis on the results, we integrate the weight of the three factors in station location selection into GIS to work out the proper number of stations by clustering analysis and get the most suitable locations for mass rapid transportation. To test and confirm the feasibility of the study method, we conduct a case study on Taoyuan Urban Area with a population of 15 million. The planning of its Urban Mass Rapid Transit System has been completed by conventional field research, resident questionnaire and all the other transit system analysis methods. The Mass Rapid Transit System locations obtained by the data mining programming of this study are very close to the planning results of the conventional methods. Thus, the method proposed in this study is practical and feasible.

1. Swift, Swift Knowledge for the Enterprise, URL: http://www.swiftknowledge.com/index.php/swiftknowledge‐for‐enterprises.html
2. IBM, Knowledge Discovery & Data Mining, URL: http://domino.research.ibm.com/comm/research.nsf/pages/r.kdd.html
3. Batty, M.,Yeh, T.(1991).The promise of Expert Systems for Urban Planning.Comput. Environ. And Urban System,15,101-108.
4. Chawla, S.,Shekhar, S.,WU, W.,Ozesmi, U.(2000).Extending Data Mining for Spatial Application：A Case Study in Predicting Nest Locations.2000 ACM SIGMOD Workshop on Research Issue in Data Mining and Knowledge Discovery
5. Ester, M.,Kriegel, H.,Sander, J.(1999).Knowledge Discovery in Spatial Databases.23rd German Conference on Artificial Intelligence
6. Fayyad, U.(ed.),Piatetsky-Shapiro(ed.),G., Smyth, P.(ed.),Uthurusamy, R.(ed.)(1996).Advances in Knowledge Discovery and Data Mining.Menlo Park, CA.:AAAI/MIT Press.
7. Koperski, K.,Han, J.(1996).Data Mining Methods for the Analysis of Large Geographic Databases.10th Annual Conference on GIS,Vancouver, Canada:
8. Koperski, Krzysztof,Han, Jiawei(1995).Discovery of Spatial Association Rules inGeographic Information Databases., Advances in Spatial Databases,.4th International Symposium, SSD'95,Portland, Maine, USA:
9. Miller, H. J.(ed.),Han, J.(ed.)(2001).Geographic Data Mining and Knowledge Discovery.London:Taylor and Francis.
10. Ng R.T.,Han J.(1994).Effective Clustering Methods for Spatial Data Mining.20th Int. Conf. on Very Large Data Bases,Santiago, Chile:
11. 王慶瑞(1999)。運輸系統規劃。Taiwan:正揚出版社。