應用分群技術於交通事故資料分析

Applying Clustering Techniques in Traffic Accidents Analysis

吳冠宏(Kuan-Hung Wu) ； 吳信宏(Hsin-Hung Wu) ； 郭廣洋(Kuang-Yang Kou)

《品質學報》 13卷3期 (2006/09) Pp. 305-312

摘要

台灣的交通雖然越來越便利，然而交通事故仍是層出不窮，時常可見車禍的發生，至今如何改善交通安全的問題仍然是個十分重要的議題。隨著時代的進步與需要，資料庫中的資料量日趨龐大，如何從資料庫中找出有用的知識以解決問題更是受到重視。本篇論文將利用資料挖掘(Data Mining)技術中的Two-Step法、自組織映射圖(Self-Organizing Maps)與K平均值演算法(K-Means Method)來分析大量的車禍資料以進行分群工作，幫助我們從其中找出有用的資訊提供建議或決策，以減少車禍的發生，其分群結果發現K-Means在車禍事故中具有比較好的分群能力，能有效的區分出不同的群集。

關鍵字

交通事故 ； 群集分析 ； 自組織映射圖 ； K平均值演算法

並列摘要

It is an essential issue in improving traffic safety in Taiwan though the traffic network has become much more convenient. In addition, the amount of data in databases has been drastically growing. Thus, gathering useful knowledge to resolve problems from databases has become critical. This paper uses the techniques of two-step method, self-organizing maps and K-means method in data mining to analyze and cluster a large amount of traffic accidents data to possibly provide useful suggestions or decisions to reduce the number of traffic accidents. The results show that K-means method performs best in clustering the data among the three techniques.

並列關鍵字

traffic accident ； cluster analysis ； self-organizing maps ； K-means method

參考文獻

Buttrey, S. E.,Karo, C.(2002).Using K-nearest neighbor classification in the leaves of a tree.Computational Statistics and Data Analysis.40,27-37.

Davidson, I.(2002).Understanding K-means non-hierarchical clustering.SUNY Albany Technical Report 02-2.(SUNY Albany Technical Report 02-2).,未出版.

Draghici, S.(2003).Data Analysis Tools for DNA Microarrays.Chapman & Hall/CRC.

Ganti, V.,Ramarkrishnan, R.,Gehrke, J.,Powell, A.,French, J.(1999).Proceedings of the 15th International Conference on Data Engineering.Sydney, Australia:

Kuo, R. J.,Ho, L. M.,Hu, C.M.(2002).Integration of self-organizing feature map and K-means algorithm for market segmentation.Computers & Operations Research.29,1475-1493.