公共自行車探索性多變量分析: 旅次特性及收費方案之影響

Abstract

公共自行車租賃系統目前於全球超過一千兩百個城市設置，多獲得普遍正面的反饋、支持，並持續於更多的城市規劃、建置。公共自行車補強了既有大眾運輸系統在第一哩路及最後一哩路的接駁、延伸，擴展其服務可及範圍、整合無縫的戶及戶旅運服務。而目前多數公共自行車租賃系統所面臨的問題是使用者欲租借時沒有自行車可借，以及欲歸還自行車時沒有車位可以停放，此供給需求不平衡的營運困難，直接地反映在系統維運效率和服務管理的問題上；於供給面，營運者大多使用小貨車對公共自行車進行調度，然而在需求面上的探討，相對仍少有較為綜觀的量化分析以針對此一問題提出更為積極的策略性管理辦法。 本研究將對於公共自行車租賃系統的需求以至於行為面建立量化的分析模式。由於個別使用者的資料難以取得，本研究擬透過巨觀層級租用交易資料分析結合租賃站點周邊環境、土地使用特性，對於需求面探討旅運需求與相關影響因素之間的連動關係，並側重於系統收費方案的影響，從而推論公共自行車租賃系統使用者的行為樣態。相對於既有文獻多針對個別站點需求進行分析，本研究將以公共自行車租賃系統站點分群起訖為基礎，探討公共自行車的旅次特性，特別是關於收費方案改變所帶來的影響。分析的資料為台北市公共自行車Youbike取消前三十分鐘騎乘免費的前後期租賃交易資料，以多變量分析探討需求面的反應。本研究利用兩個迴歸模型探討收費方案對於不同土地利用特性起訖對上旅運需求的影響；多元迴歸模型分析各分群起訖對的需求於取消前三十分鐘免費騎乘前後期的變動；有限混合模型則進一步探討收費方案的改變對於不同使用族群的影響程度差異。

Bike-sharing systems have been rapidly growing its popularity worldwide and currently established in more than 1,200 cities, which facilitates more sustainable development and fosters a trend of greener life. They provide the first-mile and the last-mile connection to complement transit service and thereby increase the use of public transport. However, one of the key difficulties in both planning and operation of a bike-sharing system manifests as the imbalance between system supply and demand. On the supply side, system operators generally use trucks to re-distribute public bikes across stations. Nevertheless, on the demand side, there are relatively few studies to holistically address this issue in the context of quantitative analysis to support strategic decision-making for more active demand management of a bike-sharing system. Hence, this research seeks to develop an analytical model to investigate the demand of a bike-sharing system and possibly provide relevant insights at a behavioral level. Due to the lack of personal data of individual users, this research explores the trip characteristics of using public bikes at the level of zonal Origin-Destination (O-D) pairs, particularly with respect to the effect of pricing scheme change, combined with the environmental factors, in terms of the land-use patterns surrounding bike-sharing stations. Based on the availability of Youbike transaction records, the bike-sharing system in Taipei, Taiwan, which contain data before and after an increased usage fee (as the cancellation of a free-usage period for the first 30-minute), the associated demand-side responses are analyzed in a multivariate context. Two regressive models are developed to investigate the effect of the change on the O-D pairs associated various land-use patterns. A multiple regression model is developed for direct before-after analysis on the variation of zonal O-D demand. A finite mixture model, on the other hand, is further constructed, which identifies three distinct components with different levels of sensitivity toward the changing price.