Mode Choice Of Low Income Commuters And The Evaluation Of Their Improvement Measures In Big Cities

With the rapid development of China’s society and economy, urban traffic problem has been more and more severe. Existing transport services are difficult to adapt to the urban development, resulting in traffic congestion. It greatly affects the normal operation of urban functions. As travel vulnerable groups, low-income people are facing more severe traffic problems. Usually, low-income groups are closely related to urban construction, services and operations. However, there is a lack of systematic scientific research in the daily activities and travel mode choice of low-income people. Their travel improvement mechanism has seldomly been proposed. Therefore, studying and solving the traffic problem of low income people is conducive to promoting the construction of a harmonious society. By analyzing the change of mode choice in different supply-demand environment, theoretical basis for traffic demand forecasting and practical application for evaluation of demand management measures will be enhanced.Based on the National Natural Science Fund Project "Mechanism between activity-travel pattern and perceived cost of low income people in big cities" and "Multidimensional travel decision mechanism of living relocation commuters’ activity chain", the research focuses on low-income commuters’ travel decision-making behavior. It includes four study levels:characteristics-mechanism-improvement measures-impact evaluation". Firstly, the thesis analyzed travel mode characteristics of low-income commuters. Then, influencing factors on mode choice were explored, focusing on the role of latent variables. Thirdly, improvement measures were proposed. Finally, the impacts of measures were evaluated by predicting travel mode choice of low income commuters under different traffic policy. Contents of the thesis include five aspects:data collection characteristics analyzing-mechanism exploration-improvement measures-prediction and evaluation.Firstly, survey program and data processing method of family unit were designed. The survey was conducted to obtain residents’ diary travel records and indicators used to construct subjective attitudes. On the basis of traditional survey method, it added personal characteristics, family characteristics and travel characteristics. Extraction procedures of analyzing trip chain information were also considered. Attitude indicators and behavioral indicators were utilized to construct latent variables. The latent variables included preferences for comfort, convenience, reliability and flexibility, safety awareness and environmental awareness.Secondly, travel mode characteristics of low income commuters were analyzed. It focused on the differences between low income commuters and non-low income commuters. Low income and non-low income people had significant differences on motorized way to travel. Low-income commuters mainly relied on public transportation. However, non-low income people relied on cars. The continuity of low-income commuters travel mode choice was higher. By analyzing travel pattern under different socio-demographics and activities characteristics, it was found that family size, private vehicle ownership, gender, occupation, license and transit IC card ownership, age, education level, trip purpose, the number of trip chain, subsistence activity duration were closely related to mode choice of low income commuters.Thirdly, attitude-behavior model were established to study influencing factors on mode choice of low income commuters from disaggregate level. MIMIC model was employed to explore family and individual variables on attitudes. The influence of family size variable on the subjective attitudes was the least significant. However, gender, driver’s license ownership and age had significant impacts on attitudes. Then, discrete choice models with and without latent variables were respectively built. By comparison of goodness-of-fit indicators, the latent variables enriched model performed better and possessed good explanation power on travel behavior. It was found that family and individual socio-demographics, activities characteristics had different effects on low income commuters’mode choice.Fourthly, improvements measures from the demand side and the supply side were proposed. First of all, based on the results of disaggregate model estimation, marginal effects and elastic theory were used to make sensitivity analysis of travel demand. Measures including the building of high-quality pedestrian environment, public transport services and cycling environment were suggested to satisfy travel needs of comfort, reliability and environmental awareness. Then, market segmentation based on attitudes was utilized to propose supply level measures aiming at travel mobility enhancement. Factor analysis was utilized to determine latent attitudes. Structural equation modeling was used to explore the causal relationship between attitudinal factors and public transit travel. K-means cluster method was employed to segment low income commuters. Travel and attitudinal characteristics of different sub-markets were examined to analyze effects of latent attitudes on public transit choices. Polices that serve differentiated public transit travel sub-markets were proposed.Fifthly, one dimensional sensitivity analysis based on support vector machine was proposed to evaluate the effects of improvement measures. By comparison of the predictive ability of SVM and MNL model, SVM showed better applicability on travel mode prediction. Then, sensitivity analysis was utilized to assess different measures. Taking mode share as a key indicator, it was found that low-income commuters’travel mode choices changed quite differently under different traffic policies. Finally, for the purpose of "green mode share maximization" and "public transportation mode share maximization", three important measures were found to be useful, including improving transit IC card ownership, improving travel reliability and improving the environmental awareness of low income commuters.Conclusions of the thesis will add to the rich body of activity-travel pattern study, especially for low income commuters, and enhance the theoretical basis of travel demand forecasting. It is hoped that transportation planners pay more attention to low income groups when making urban transport development strategy. The study may also provide theoretical support for optimization of traffic resources allocation and establishment of sound mechanism to better meet low income residents’travel needs.