應用資料探勘於交通事故環境之關聯規則與預測

Using Data Mining to Associating Rules and Predictions on the Traffic Accident Environment

張敏亮(Min-Liang Chang)

指導教授 ： 吳信宏

亞洲大學/資訊學院/資訊科學與應用學系碩士班/碩士(2005年)

摘要

依據內政部警政署的統計，機動車輛所造成交通事故一直是這幾年台灣死因的前幾名，所以顯而昜見的車禍肇事意外已經嚴重影嚮到國民的生命財產之安全與保障。因此如何降低交通事故發生率，讓人民財產損失減少，生命獲得保障儼然是目前最重要的課題之一。

隨著近年來資料探勘技術的發展，其應用在各個學門都有良好的成效，因此本研究將針對交通事故的現場環境，利用資料探勘技術來挖掘當中隱含的資訊，希望能提供給相關部門，對於交通工程、道路環境設計或建造時一個參考，以期望能使得肇事發生率能獲得有效的改善，降低交通事故對於人民生命財產的威脅。

基於以上因素，本研究利用關聯規則方式來找尋造成車禍之環境成因，以及運用決策樹分析能提供明顯分類特徵，並且能夠支援數值型及類別型資料之特色，針對交通事故環境及車禍間關係來發展預測模型，並以88及89年度車禍記錄作為訓練資料，90年度車禍記錄作為測試資料，其所得結果大部份能有6成以上預測準確度。此預測模型圖能提供作為檢示道路危險程度之依據，也能作為開發交通事故專家系統之基礎。

關鍵字

資料探勘 ； 關聯規則 ； 決策樹 ； 分類迴歸樹

並列摘要

Traffic accidents are one of the critical reasons to cause deaths in Taiwan based upon the Interior Department analysis. One of the most important tasks for our government is to reduce the number of traffic accidents as well as to protect the lives of people in Taiwan. As data mining techniques have been well developed and widely and successfully applied in many areas, this study uses data mining techniques to discover the hidden information in the raw data to provide useful information as a reference by improving the traffic environment.
According to the above discussions, this study finds out the reasons that result in traffic accidents by association rules. Moreover, decision trees are applied to provide a clear indication of which fields are most important for prediction or classification as well as to handle both continuous and categorized variables. The relation between the traffic environments and traffic accidents can be used to develop the prediction model that leads more than 60% of accuracy in prediction. Finally, the road hazardous degrees can be defined and the basic expert system for traffic accident analysis can be developed based upon the prediction model.

並列關鍵字

data mining ； association rule ； decision tree ； classification and regression tree

參考文獻

10. 鍾明璇，「應用關聯規則技術有效輔助以向量空間模型為基礎之文件群集法」，中原大學資訊管理學系碩士論文，2002。

12. Al-Ghamdi, A.S., “Using logistic regression to estimate the influence of accident factors on accident severity,”Accident Analysis and Prevention, 34, pp. 729-741, 2002.

17. Dyche, J., The CRM Handbook: A Business Guide to Customer Relationship Management, Addison-Wesley, 2002.

18. Frawley, W. J., G. Paitesky-Shapiro, and C. J. Matheus, Knowledge Discovery in Databases: An Overview, AAAI/MIT Press, 1991.

19. Grupe, F. H. and M. M. Owrang, “Data mining discovering new knowledge and cooperative advantage,” Information Systems Management, 12(4), pp. 26-31, 1995.