透過您的圖書館登入
IP:3.235.249.219
  • 學位論文

資料—視覺—判釋方法於公車資料分析

Data-Visual-Judgment Approach for the Analysis of Bus Data

指導教授 : 康仕仲

摘要


公車路線重疊會間接導致公車連班以及誤點等現象,造成服務品質下降。隨著科技的進步,現在已經累積大量公車相關的資料,包括公車動態資料以及民眾使用電子票證的資料。傳統的交通分析方法正面臨高維度巨量資料的挑戰,它們無法有效的發掘區域性的偶發事件。在這項研究中,我們提出了一個資料-視覺-判釋的方法,此方法包括三個步驟,依序為建構資料模型(Data Modeling),視覺探索(Visual Exploration),人為判釋(Human Judgment)。在第一個步驟中,本研究從公車動態系統和公車電子票證資料庫提取原始資料,進行資料清理及資料處理後,建構資料模型。在第二個步驟,我們開發一個視覺探索工具,由公車分時堆疊圖(Hourly Bus Stacked View)以及時空圖(Time-space View)組成的圖型介面互動工具。第三步驟是建立一個人為判釋資料程序,使用者可依循這個程序從視覺探索工具中歸納出低乘載量和過度擁擠的巴士出現的時段。運輸業者透過這個方法找到高乘載量時段及低營運效率時段之後,可以針對供給不足的時段研擬對策,另一方面,可以考慮擴大低載客率時段的公車班距。為了驗證方法的有效性,本研究使用2015年4月期間搜集的860萬筆悠遊卡資料和380萬筆公車動態資料,選擇台中市豐原到東勢之間的路廊作為研究對象。本研究設計了三個使用者測試題目,對21個受試者進行了測試,受試者分別使用本研究提出的視覺探索工具以及傳統的時空圖來回答問題,結果顯示,在三個問題中,受測者使用視覺探索工具答題相較於時空圖節省時間,另外,受測者使用視覺探索工具答題的正確率比使用時空圖高。因此,本研究認為資料-視覺-判釋方法未來能幫助交通業者更有效率的發掘隱藏在公車系統中的問題,往後本研究的成果將可融入交通規劃的流程,以輔助公車運輸的決策。

並列摘要


The explosion in the volume of available bus data has posed a challenge to using traditional statistical approaches for analysis. It is difficult to identify low-frequency events such as the occurrence of near-empty buses and overcrowded buses from the enormous amount of data. Especially for overlapping bus routes, the interrelationship between bus routes and the dynamics of travel behavior increase the difficulty of bus data analysis to some extent. In this research, we proposed a Data-Visual-Judgment method (D-V-J method) to facilitate stakeholders in finding irregular bus services using bus data. D-V-J method includes three steps: data modeling, visual exploration, and human judgment. In the first step, we extracted the raw data from an Automatic Vehicle Location and Automatic Fare Collection database, then these data were processed and integrated into a single dataset providing bus supply and passenger load information. In the second step, we developed a visual explorer comprising an hourly bus stacked view and time-space view. The third step is to establish a procedure for users to judge near-empty buses and overcrowded buses directly from the hourly bus stacked view and address the cause of these irregular bus services through exploring the time-space view, then they are able to investigate the improvement of bus operation policies accordingly. To validate the effectiveness of the D-V-J method, we conducted a user test with 21 subjects by using 8.7 million EasyCard transactions and 3.8 million bus log data collected in April 2015 and selected Fengyuag-Dongshih pathway in Taichung city as our target bus corridor. Each subject performed the judgment tasks and answered three questions regarding practical bus operation by employing both the proposed visual explorer and the conventional method (time-space diagram). The results showed that using the visual explorer to perform the tasks and answer the questions involves shorter completion times and have higher success rates than the conventional method, providing evidence that applying D-V-J method can improve the efficiency of identifying irregular bus services in high-dimensional bus data. Hence, this research will benefit the transport administration on governing service quality and also the bus operators on maintaining service reliability.

參考文獻


Bartholdi, J. J., and Eisenstein, D. D. (2012). “A self-coordinating bus route to resist bus bunching.” Transportation Research Part B: Methodological, 46(4), 481–491.
Chen, W., Guo, F., and Wang, F. (2015). “A Survey of Traffic Data Visualization.” IEEE Transactions on Intelligent Transportation Systems, 16(6), 2970–2984.
Chey, J., and Holzman, P. S. (1997). “Perceptual organization in schizophrenia: Utilization of the Gestalt principles.” Journal of Abnormal Psychology, American Psychological Association, 106(4), 530–538.
Cook, K., Earnshaw, R., and Stasko, J. (2007). “Guest Editors’ Introduction: Discovering the Unexpected.” IEEE Computer Graphics and Applications, 27(5), 15–19.
Hsieh, O. (2016). “Human computer interaction and data visualization.” Advanced Writing: Pop Culture Intersections.

延伸閱讀