Study On Commuter Activity-travel Behavior Analysis

Traffic congestion and subsequent social and environmental problems have becomethe bottlenecks for sustainable urban development. The main cause is the imbalance on thedemand and supply of urban traffic system. Therefore, there’s increasing interest in thedevelopment and implementation of travel demand management (TDM) strategies. Thesestrategies are aimed at effectively managing and distributing travel demand in both spatialand temporal dimensions.Travel behavior analysis is one of the frontiers of transportation science. It issupported that a systematic and comprehensive study on traveler behavior analysis isessential for the diagnosis and treatment of urban traffic problems and the scientificplanning and implementation of TDM measures. Activity-based approach, whichexplicitly recognizes that travel demand is derived from the need to pursue activities, is amost promising alternative to the current travel behavior analysis methodology. In the pasttwo decades great progress has been made in activity-based travel forecasting and TDMstrategy evaluation both at home and abroad. But research on the joint and causalrelationships among multiple endogenous variables, which plays an important role in theunderstanding of travel decision mechanism, is relatively insufficient. Some studies in thisfield have already been carried out in developed countries but fewer insights are available in the literature in the context of developing countries. However, the characteristics ofeconomy level, land use, household socio-demographics, travel and activity attributes andtraffic system service level in China are all different from those of the western countriesand it is necessary to study the interrelationships among traveler activity and travelattributes conforming to the local situation.The analysis of commuter travel behavior is of great importance since it has a decisiveimpact on peak-period congestion, which is now the foremost transportation problem.Motivated by the relevance of the subject in the current context, the focus of this researchis to present a comprehensive analysis of commuter activity-travel attributes and theinterrelationship among them based on the processing and analysis of household travelsurvey data, which may shed considerable light on travel demand forecasting and TDMmeasure evaluation. Results from this study will expand and enrich the theoreticalframework and analysis methods of travel behavior analysis and contribute to theestablishment and carrying out of high performance TDM measures by presentingscientific and effective decision analysis methods. The main research contents aresummarized as follows:(1) Guided by the data requirement of activity-based travel behavior analysis, dataprocessing method including rules about trip data validity judgment and data arrangementis presented according to the characteristics of traditional trip-based travel survey data.Then travel mode and interchange pattern, characteristics of travel time and the timing andduration of activity episodes are examined. The trip-based travel data are converted to tripchaining records composed of activities, stops and trips, which are used as the basic database for the following research.(2) Urban traffic problems are typically characterized by peak-period trafficcongestion during commuting time thus the research of commute travel mode choice canprovide practical tools for the diagnosis of urban transportation problems and theregulation and control of commute travel demand. The present study is aimed to provide anew method addressing some of the shortcomings of existing researches by taking thefollowing factors on mode choice into consideration: alternative means of choice setrepresentation, reasonable model parameter identification and the improvement based onthorough analysis of forecasting results. The validity of the provided method in regard to policy and planning scenarios is confirmed by the analysis of the mode shift under anincrease in vehicle ownership.(3) As commute trip chains become more complex, flexibility of travel mode becomesmore important and the mobility service offered by public transport is less attractive. Inorder to obtain a better understanding of the impact of public transport on commute tripchain, this paper investigates the interaction between commute mode choice and thecomplexity of trip chaining pattern. The relationship between the two choices isrepresented in a recursive simultaneous bivariate probit model by the introduction ofdummy variables indicating choice results and the correlation coefficient between the tworandom errors. Then the direct and indirect effects of individual and householdsocio-economic attributes, the characteristics of work activity and the trip chaining patternon the choice of commute travel mode are examined based on the analysis of the marginaleffects of a series of discrete and continuous explanatory variables. And the importantimplications of the research findings for travel demand forecasting and management arediscussed.(4) Travel demand management strategies and transportation control measures areinherently linked to the time dimension. People’s time allocation patterns, havingprofound impact on the daily activity and travel arrangement, are the research focus oftravel behavior analysis. This dissertation first applies the fractional split distributionmodel to investigate commuters’ allocation of time to in-home and out-of-home non-workactivities and demonstrates the applicability of the model for determining activity andtravel behavior adjustments through numerical situations of compressed work week andflexible working hours. Then the duration model approach which integrates the notion ofthe temporal dynamics is applied to the analysis of commuters’ daily travel time cost. Therelationships between daily travel time and socio-economic attributes and activity andtravel characteristics are analyzed and the time allocation pattern is partly represented bythe derivative and competing relationships of activity and travel.(5) Commuters often introduce non-work activities to the basic home-work-homechain because of the temporal-spatial constraints from work activity. This commute tripchaining may explain the rise in non-work trips occurring in peak periods and is posited asone reason for increased congestion problem in peak periods. Therefore, this dissertation focuses on workers’ decisions of activity participation and timing. Taking non-workactivity timing data as a censored sample a censored probit model is established to answerthe following questions: what influences the decision to participate in a non-work activityand among those who pursue non-work activities what contributes to the chaining of theactivity to work. The model consists of the selection equation and the outcome equation,and the interrelationship between activity participation and timing is represented as thecorrelation between the two random errors. The model provides a better understanding ofhow workers make non-work activity decision in relation to work, which is a majorrequisite to improving the performance of travel demand modeling as well as thedevelopment of congestion relief policies.(6) Members in a household share various household resources such as income, livingspace and transportation tools and play different roles in the household. As a result,household members interact in daily activity-travel choices. A structural equation model isused to explain household allocation of activity and travel between household heads. Themodel attempts to capture the substitution, companion and complementary relationshipsinvolving both activities and the associated travel both within and between the twomembers. Since the trip to work is a reflection of location decisions made regarding bothhousing and employment locations by all members of a household there may be complexinterrelationships among individual household members. A bivariate ordered probit modelis applied to analyze the interrelationships between spousal commuting decisions. Inparticular it is further explored whether spousal trips to work appear to be substitutes orcomplements for one another.The main contributions of this dissertation research are as follows:(1) Aimed at addressing some of the shortcomings of current mode choice models thispaper investigates alternative choice set representation and reasonable model identification.It is demonstrated that an explicit model of unavailability of some alternatives provides abetter fit and captive effects are observed among commuter segmentations. A predictedoutcome matrix is constructed to distinguish those aspects of travel behavior that are notwell-captured by the model. This process is designed to improve the performance of themodel.(2) The relationship between the use of public transport and the complexity of commute trip chaining pattern is analyzed in a recursive simultaneous bivariate probitmodel, which provides a powerful analysis tool for the examination of interactions andmutual restrictions between two discrete choices. Research findings from this effort haveimportant implications for the development of activity-based travel forecasting systemsand for the understanding of travel decision process.(3) Commuters’ daily travel time is analysed in a hazard duration model, whichenriches the quantitative analysis method of travel time and is of great importance for theprecise understanding of commuters’ time allocation behavior. It models the conditionalprobability of the end-of-duration of travel, given that it has lasted to a specified time andpermits the likelihood of ending to be dependent on the length of elapsed time. Theresearch results demonstrate that the behavior of most commuters can be represented bytravel time minimization mechanism.(4) The participation and timing of non-work activities of commuters are analyzed bycensored probit model, where the non-work activity choice data are taken as censoredsample. The model is composed of two equations: activity participation is analyzed in theselection equation, activity timing is analyzed in the outcome equation and theinterrelation between participation and timing is accounted by the correlation between thetwo random errors. This provides an exploration to extend probit model for censoredsample analysis in activity-travel behavior analysis.(5) Models of activity engagement and time allocation among household heads aredeveloped and estimated in order to identify the trade-offs and complementaryrelationships among household members’ activity and travel engagement patterns. Ingeneral it is an attempt for the understanding of the intra-person and inter-personinteractions in the context of non-work activity and travel patterns and provides basis forfurther studies of various interactions among all household members simultaneously.