共享單車系統顧客抵達分布分析

An analysis of patron arrival distributions of a bike sharing system

施宛廷(Wan-Ting Shih)

指導教授 ： 周雍強

國立臺灣大學/工學院/工業工程學研究所/碩士(2018年)

https://doi.org/10.6342/NTU201800654

摘要

近年來，為了節能減碳與降低環境汙染，世界各國都積極建置共享車系統。自行車本身具有降低能源消耗、減少污染，以及有助於紓緩交通的特性，搭配新興的商業模式「共享經濟」，變成具有甲地租乙地還或隨租隨還的服務系統，解決民眾搭乘交通運輸後「最後一哩(Last Mile)」的問題，並讓城市的交通路網更完整。
在共享單車的營運期間，為了避免顧客遇到無車可借或無位可還的問題，營運者會派遣調度人員開調度車進行調度，使站點車輛的供需平衡。雖然調度可讓缺車缺位的問題改善，但需求的不確定性仍會造成缺車缺位出現，因此，要將調度模式優化，站點的使用需求預測是必須的。
在建構需求預測模型時，須考量顧客抵達的分布，也就是需求出現的頻率。大部分的研究根據理論，將借車與還車的頻率設為Poisson分配，然而，實際情形可能不是如此，因此，本研究透過觀察站點，蒐集實際的使用者數據，分析現行的站點營運狀況，在考量不同雨量的影響下，利用卡方檢定法，對借車與還車的使用頻率進行適合度檢定，分析結果在分別的雨量影響下，借車與還車的使用頻率幾乎不拒絕Poisson分配的假設。

關鍵字

共享單車 ； 顧客抵達分布 ； 站點分析 ； Poisson分配 ； 適合度檢定

並列摘要

In recent years, in order to save energy and reduce carbon emissions and environmental pollution, all countries in the world are building bike sharing systems actively. The benefits of Public Bicycle-sharing System (PBS) are reducing energy consumption, pollution and helping to reduce traffic. When PBS linked to the latest business model of "sharing economy" by turning itself into a service system, it provides other users a chance to rent and return in another places, make the city's transportation network more efficiency and complete, and the problems of "Last Mile" is also solved.
For better operating the PBS, operators need to send staffs driving trucks or motorbikes to dispatch the bicycles from one location to another as to balance the demand and supply of bicycles on the site. Although the dispatching can make the problem of the lack of bicycles and the uncertainty of demand reduced, it is still necessary to optimize the scheduling mode and forecast the demand of enhance the bicycles utilization.
In the construction of PBS demand forecasting model, we need to consider the distribution of customer arrivals, which is the frequency of demand. Based on current theories, they tried to set the frequency of patron arrival as the Poisson distribution. However, this may not be the case, in order to prove the Poisson distribution, this thesis choses the approach by looking at one site to collect actual user data, analyzing the patron arrival under the influence of different rainfalls, and using chi-square test to test the suitability of the frequency of demand. The result of chi-square test almost based on the assumptions of patron arrival as Poisson distribution.
In conclusion, my study reflects that renting patron arrival under the different rainfalls can be assumpted as Poisson distribution, but the returning patron arrival except no rain situation can be assumpted as Poisson distribution.

並列關鍵字

Bike sharing ； Patron arrival distribution ； Station Analysis ； Poisson distribution ； Test of Goodness-of-Fit

參考文獻

[3] 王俊偉(2011)，以系統模擬探討公共自行車租借系統之建置及營運策略，成功大學資訊管理研究所學位論文

[9] 姚漢禱 (1994)，波松模式(Poisson Model)在運動測驗上的理論和應用，中華民國體育學會體育學報，第十七輯，頁159-168.

[18] 解鴻年、張馨文(2011)，新竹科學城民眾使用公共自行車意願分析，建築與規劃學報，第十二卷，第三期，頁237-254

[20] 劉俊志、吳宜珍、江介倫、鄭克聲(2007)，線性動差比適合度之檢定力測試，農業工程學報，第53卷，第4期，頁80-91.

[14] 張書瑋(2017)，共享經濟 傳統與創新的戰爭，會計研究月刊，第378期，頁58