An Assessment of The Effect of Contraflow Bus Lane on Ridership Transportation Planning Journal Vol. 26 No. 3 September 1997 PP.577~592 運輸計畫季刊 第二十六卷 第三期 民國八十六年九月 頁577~頁592 張學孔 Shyue Koong Chang 陳信雄 Hsin-Hsiung Chen

自我介紹 蕭復元 南投、台中 學歷 實習工讀 興趣 成大交管系 交大運管所物流組碩一 任老師 運輸與行銷管理研究室 台鐵台中站 中華電信台中營運處 興趣 棒壘球、唱歌

文章出處 運輸計劃季刊 交通部運輸研究所 http://www.iot.gov.tw/Modules/Quarterly/Quarterly-List?node=ac53d790-78a9-490f-b3a2-df957b99764d&c=1

作者簡介 張學孔 陳信雄 國立台灣大學 土木工程學系 交通組 教授 研究領域 國立台灣大學 土木工程學系 交通組 博士 大眾運輸 運輸經濟 智慧運輸系統 陳信雄 國立台灣大學 土木工程學系 交通組 博士 台灣高鐵公司行銷處 協理

大綱 Introduction Problem for Time Dependent Data Intervention Analysis Data Analysis Forecasting Conclusions and Recommendation

Introduction

逆向公車專用道 運輸系統管理 TSM 1990年2月啟用 信義路、仁愛路單向車道 仁愛路 信義路 杭州南路 敦化南路 INTRODUCTION 逆向公車專用道 運輸系統管理 TSM 1990年2月啟用 信義路、仁愛路單向車道 杭州南路 敦化南路 仁愛路 TDMTransportation Demand Management運輸需求管理是使用各種誘導措施以鼓勵使用大眾運輸、腳踏車及走路並同時使用各種限制措施，以抑制一人一車的旅次同時用非運輸手段轉移旅運時間以有效率使用現有的運輸設施 TSMTransportation System Management運輸系統管理是一種短期性低成本的運輸系統改善策略在現有的道路及大眾運輸設施的基礎上以有效率的經營管理來提高運輸服務水準、節約能源、減少污染及交通安全等運輸目標 仁愛路單行道段 東向逆 信義路西向逆 信義路

INTRODUCTION 逆向公車專用道

Problem for time dependent data

Effect of Event Over Time PROBLEM FOR TIME DEPENDENT DATA Effect of Event Over Time Conventional Compare between Q1 and Q3 Result: increase Δq1+Δq2 Actual Δq1 → time-dependent factors Result: increase Δq2

INTERVENTION ANALYSIS

Time Series Method An+1, An+2, An+3, …, An+m Cn+1, Cn+2, Cn+3, …, Cn+m INTERVENTION ANALYSIS Time Series Method 1 Obtain n data B1, B2, B3, …, Bn 2 Assume the event will not occur and forecast next m periods An+1, An+2, An+3, …, An+m 3 Obtain m actual time series data Cn+1, Cn+2, Cn+3, …, Cn+m 4 Compare

DATA ANALYSIS

大都會客運22路-信義路 1987.01 1990.02 1994.10 Before: 37 After: 57 逆向公車專用道啟用 DATA ANALYSIS 大都會客運22路-信義路 Before: 37 After: 57 1987.01 1990.02 1994.10 逆向公車專用道啟用

大都會客運22路-Ridership Trend DATA ANALYSIS 大都會客運22路-Ridership Trend Daily Monthly February Winter break & Chinese New Year Less days

FORECASTING

Box-Jenkins ARIMA model FORECASTING Box-Jenkins ARIMA model ARIMA (Autoregressive Integrate Moving Average) The general model: p: Parameter of Autoregressive (AR) Model q: Parameter of Moving Average (MA) Model d: Degree of Difference P: Parameter of Seasonal Autoregressive (SAR) Model Q: Parameter of Seasonal Moving Average (SMA) Model D: Degree of seasonal differene s: Length of seasonal cycle ARIMA(p,d,q)(P,D,Q ) s

Monthly Ridership ARIMA(1,0,0)(1,0,0 ) 12 FORECASTING Monthly Ridership ARIMA(1,0,0)(1,0,0 ) 12 Y t =0.87 Y t−1 +0.84 Y t−12 + 0.87 0.84 Y t−13 +10451.5 (12.04) (13.35) (9.09) R 2 =0.75 85% Confidence Interval 95% Confidence Interval

Daily Ridership ARIMA(1,0,0)(1,0,0 ) 12 FORECASTING Daily Ridership ARIMA(1,0,0)(1,0,0 ) 12 Y t =0.879 Y t−1 +0.769 Y t−12 + 0.879 0.769 Y t−13 +467.86 (12.95) (8.64) (8.64) R 2 =0.72 95% Confidence Interval 85% Confidence Interval

Monthly Comparison-Observed & Forecast FORECASTING Monthly Comparison-Observed & Forecast 85% Confidence Interval 95% Confidence Interval

Daily Comparison-Observed & Forecast FORECASTING Daily Comparison-Observed & Forecast 85% Confidence Interval 95% Confidence Interval

Fisher Sign Test Approach FORECASTING Fisher Sign Test Approach Nonparametric test Difference between two series of data before March 1992 (95% C.I.) July 1993 (85% C.I.) H 0 : There is no significant effect on ridership H 1 : Not accept H 0 e.g. Monthly comparison (95% C.I.) 26 differences 20 (+) i=0 6 B(26,i,0.5)≅0.007 符號檢定（The Sign Test） 主要是用來檢定兩個相依樣本的差異性，其方法是將觀測值分為三類：若X＜Y則歸入（＋）類，若X＞Y則歸入（－）類，若X＝Y則歸入同分（tie）類，以進行檢定。 ∵0.007<0.05 ∴ Reject H 0 The ridership significantly increase

FORECASTING Ridership Comparison

Conclusion and recommendations

CONCLUSIONS AND RECOMMENDATIONS Box-Jenkins ARIMA model accurately forecast the ridership If the contraflow lane was not implemented The conventional method underestimate the effects since the time variation was neglected The ridership increased 29%

Recommendations Multivariate time series model CONCLUSIONS AND RECOMMENDATIONS Recommendations Multivariate time series model Effects of the contraflow bus lanes on bus operation characteristics Travel speed Service frequency Operating cost Network effects of the exclusive bus lane on bus operation and travel characteristics of Taipei City

The End Thank you