交通資訊影響下之用路人路徑選擇行為：累計期望理論模型之構建與探討

Road users’ Routing Choices under the Influence of Traffic Information: Development and Investigation of Cumulative Prospect Theory

田宜欣(Yi-Hsin Tien)

指導教授 ： 董啟崇

淡江大學/商管學院/運輸管理學系碩士班/碩士(2015年)

https://doi.org/10.6846/TKU.2015.00276

摘要

基於路網車流的不確定變化特質，交通管理者所提供的路況資訊亦需反應路況之不確定狀態；另一方面，用路人對不確定路況資訊的反應也因內在特性與自身經驗影響路徑決策。在探討不確定狀態下之決策行為理論中，以風險觀念為基礎之期望理論發展於1979年，最初應用於探討投資、金融等領域，之後於1992年修正為累計期望理論。近年來續見應用累計期望理論探討用路人之路徑決策行為，即將路況資訊可能出象(outcome)(如：旅行時間)與附屬機率之形式呈現，並獲得若干成果。然檢視累計期望理論目前應用於交通範疇中仍存在若干模式建構與研究實驗設計上須考量之課題，本研究探討課題包含：動態參考點設定、動態路徑決策影響、延伸有限的資訊出象狀態、替代方案(路徑)的期望值出象差異狀態等以及以反應用路人之風險態度分群。而本研究採用控制實驗法，透過靜態問卷與動態控制實驗兩階段分別進行路徑決策行為調查，控制實驗主要以兩個層面進行調查：其一為控制資訊出象的程度多寡，另一為控制路徑方案的期望值、旅行時間、附屬機率資訊，以探討累計期望理論中用路人面臨不確定資訊之決策行為與風險態度。
本研究工作內容包含三部分: (1)初始問卷調查用路人基本特性及對不確定之出象強度資訊給予的決策表現。(2)擬真動態控制實驗，設計動態資訊導引擬真環境，觀測與探討受測者在兩種不同完整程度的出象強度資訊導引下的決策行為，並以累計期望理論為基礎建構與校估參數。(3)系統化比較兩種不同程度之出象強度下之模式解釋能力，以及風險態度表現與參考點變動。
綜整本研究實證結果之重要結論：(1)本研究應用累計期望理論分別設計兩項出象強度資訊與三項出象強度資訊之動態控制實驗，發現用路人在不同程度之出象強度資訊給予下，其風險態度表現相似，顯示出象強度資訊完整度給予差異對決策行為影響程度較小。(2)本研究以絕對偏差均數作為衡量指標計算不同出象強度資訊下的累計期望模式解釋能力，發現給予兩項出象強度資訊之決策行為，累計期望模式具有較佳的解釋能力。(3)觀察靜態問卷與動態控制實驗的實證結果差異，發現於靜態問卷調查下較傾向風險規避態度，於動態控制實驗下較傾向風險追求，可發現用路人於決策過程中並非保持一致的決策標準。(4)在控制期望值、旅行時間及附屬機率資訊給予的情境組合下，可發現用路人的決策行為會顯著的受到附屬機率資訊影響。

關鍵字

累計期望模式 ； 參考點 ； 風險態度 ； 交通資訊不確定性 ； 控制實驗 ； 路徑選擇

並列摘要

Based on uncertain variation of traffic flow of network, traffic information provided by traffic managers need to reflect uncertain traffic status in network. On the other hand, drivers’ route choice behavior toward uncertain traffic information are affected by their essential properties and experiences themselves. One of choice behavior theories which with risk concept—cumulative prospect theory (CPT) is developed initially in 1979, were applied originally to research investment, finance related fields, in recent years CPT are applied to survey drivers’ route choice behavior, which traffic information described by possible outcome(eg.travel time and correspond possibilities), and some studies reached achievements. However, some model construction and experimental design of CPT on traffic issues were founded while viewing applied CPT traffic studies, and this study mainly research traffic issues of CPT including: setting of dynamic reference point, dynamic route choice effect included, extending limited information outcome status, expected values of outcome variation status of route alternatives, and risk attitude grouping depend on decision behavior characteristics. This study adopted controlled experiment method, conducting route decision behavior by a static questionnaire and a dynamic experiment respectively, controlling two contexts experimental design: one was controlling completion of travel time information of routes, the other was controlling differences and similarities of expected value, travel time and corresponded possibilities information of route alternatives, which to survey drivers’ route choice behavior and risk attitude when encountering uncertain information based on cumulative prospect theory.
The tasks of this study were in three folds: the first task was a questionnaire to investigate subjects’ basic information and decision behavior toward obtaining uncertain outcome probabilities information. The second task was a verisimilitude dynamic controlled experiment, by designing dynamic information guidance surroundings to measure and research subjects’ route choice behavior on two different completion level of travel time information, and construct model by CPT and calibrated parameters. The third task was conducting a systematic comparison two different level of complete outcome probability information of CPT explain abilities, risk attitude and the reference point changes.
Conclude the main findings in this study are summarize in the following:
1. Respectively design two outcome probability information and three outcome probability information to construct dynamic controlled experiment based on CPT, finding drivers’ risk attitude was similar, which means drivers’ decision behavior were less affected by complete level of traffic information.
2. Trough mean absolute deviation to measure model explain ability under different information, we found the explain abilities of giving two outcome probabilities information based on CPT is best.
3.Campared the questionnaire results and dynamic controlled experiment results, people tend to risk averse attitude in questionnaire results, on the other hand, people tend to risk seeking attitude in dynamic controlled experiment results, hence driver didn’t keep consistency of decision standard.
4. Drivers’ route choice behavior is obvious affected by possibilities in traffic information under controlling expected mean, travel time and correspond probabilities.

並列關鍵字

Cumulative Prospect Theory ； Reference Point ； Risk Attitude ； Information Uncertainty ； Controlled Experiment ； Route Choice

參考文獻

40. 蘇秋如(2006)，「個人化路徑導引資訊：動態路徑選擇行為模式之應用」，私立淡江大學運輸科學研究所碩士論文。

1. Avineri, E. (2006). The effect of reference point on stochastic network equilibrium. Transportation Science, 40(4), 409-420.

2. Ben-Elia, E., Erev, I., and Shiftan, Y. (2008), “The combined effect of information and experience on drivers’ route-choice behavior,” Transportation, Vol.35 No.2, pp.165-177.

3. Ben-Elia, E., and Shiftan, Y. (2010), “Which road do I take? A learning-based model of route choice with real-time information,” Transportation Research Part A, Vol.44, pp.249-264.

4. Cohen, M., Jaffray, J. Y., and Said, T. (1987),” Experimental comparison of individual behavior under risk and under uncertainty for gains and for losses,”Organizational behavior and human decision processes, Vol.39, issue1, pp.1-22.