利用資料探勘探討時間對車流量大小因素之研究- 以后里收費站為例

Applying Data Mining Technology for the factor of running of car about time –Illustrated by the case of toll station in Houli

陳炎長(Yan-Chang Chen)

指導教授 ： 陳大正

國立虎尾科技大學/管理學院/資訊管理研究所/碩士(2011年)

https://doi.org/10.6827/NFU.2011.00076

摘要

高速公路收費站採用計次道路收費之技術已經成功導入相當長的一段時間。但由於經濟起飛，台灣地區每年汽車以二十萬輛快速增加與成長，影響到行車品質。透過收費來降低短程旅次，已經成為世界各國改善交通的普遍做法。然而，逢年過節，各收費站前仍常有長龍壅塞之狀況，造成用路人的困擾。
為了改善車流因繳費而壅塞之狀況，本研究從時間因素對車流量大小進行研究，探討影響民眾經由收費系統過站壅塞因素，以行為理論作為解釋民眾行為的理論基礎。本研究以資料探勘技術進行分析，資料共蒐集六個時間相關多樣變數和屬性類型，使用SPSS 軟體中的決策樹來進行資料的分析。
本研究在「刪修重要性」設定為75%，「子分支最少記錄數」使用預測值2，依據高公局后里收費站南北向總筆數做分析，經分析研究結果其準確率均高達94%以上，有此可證明預測正確。最後藉由資料探勘技術中的決策樹演算法，分析週休二日、平常日與連續假期之后里收費站車流量分析，研究結果顯示該站車流量“ 高流量 ”與 “低流量”與週休二日、平日與連續假期各尖峰時段、離峰時段有絕對性的關係。

關鍵字

資料探勘 ； 車流量 ； 人工收費 ； 決策樹

並列摘要

Conventional fare collection system has been successfully applied to freeway toll station for a long period. Because of the rapid development of the econmy in Taiwan, the amount of cars grows by 200 thousands per year. It deeply affects the traffic quality. Decrease the vehicle frequency by fare collection has been broadly adoped to improve the flow by other countries. However, in holidays, traffic congestion still occurs in the toll station areas.
In order to improve the traffic jams caused by fee collection, this study tries to realize the relationship between time factor and vehicle flow. Discuss the the reasons of the traffic congestion hapeend in fee collection stations by behaviorual theory.
During the process, we set the value of attribute “the importance of deleting and modifying” to 75%, as well as the value of “Minimum records per child branch” to 2. The historial data is derived from bi-directional vehicle flow of Hou-Li toll station. The accuracy rate of the test result is higher than 94%, which can support our hypothesis and predictions. Finally, we also perform the decision-tree algorithm to examine the corresponding flows in weekends, weekdays and consecutive holidays. The results demonstrate that traffic flow has definitely relationship with the type of holidays.

並列關鍵字

Data Mining ； Traffic Flows ； manual charge ； Dcision Tee

參考文獻

13.黃益三【1997】，高速公路動態交通時間狀態特性之分析、預測與應用，國立交通大學交通運輸研究所博士論文，民國86 年。

19.陳琇宜【2009】，應用資料探勘於全額交割股恢復交易之探討，國立虎尾科技大學資訊管理系碩士論文。

20.許佳豪【2010】，運用資料探勘技術建構電子業財務危機預警模型，國立虎尾科技大學資訊管理系碩士論文。

28.Brownstone, D., and Ghosh, A., “Drivers’ Willingness-to-pay to Reduce Travel Time: Evidence from the San Diego I-15 Congestion Pricing Project,” Transportation Research Part A, Vol. 37, pp. 373-387, May, 2003.

29.Assessing Expected Accuracy of Probe Vehicle Travel Time Report, Journal of Transportation Engineering, 1999.