Modeling Analysis The Travel Choice Behavior Based On Mental Accounting

It is important way to analyze the feature of traveler' choice behavior that is using stated preference data to estimate travel choice behavior model and Data quality and model performance are two key factors which could impact the accuracy of analysis results.At present, Discrete Choice Model and Expected Utility Model are used most commonly in travel choice behavior. The two kinds of model assume that people adopt rational decision-making principle when they make a choice, but the assumption is not reasonable. In fact, people are bounded rationality when they make a decision. Although Prospect Theory expands the hypothesis and suggests people is bounded rationality when they make a choice. However, Prospect Theory cannot describe the decision-making problem with multidimensional attributes,and it cannot explain why the choice results will change under different decision-making circumstance. Prospect Theory is limited to apply in the travel behavior filed because of these disadvantages. In order to overcome the disadvantages of Prospect Theory and to expand rational decision-making principle, the travel choice behavior model would be developed based on Mental Accounting in this paper. At present, Mental Accounting is almost used to analyze the consumer choice behavior and investing choice behavior. It is still in the stage of theoretical analysis in travel choice behavior, there are some aspects should be expanded constantly, such as, the decision-making principle, the value function form, the parameter estimation algorithm and the model development.Firstly, it's demonstrated that when people make travel descison they will adopt the topical accounts spontaneously by Analysis of Variance based on the survey data, it can explain the situation that the choice will change if the decision environment is different. Then, the value function was developed in travel choice behavior and the new way of setting reference point also was proposed. Meanwhile,we found there is significant preference difference between time and money.People are risk aversion when they face the gain and loss of time attribute, but when people face the gain of money they are risk aversion, and if people face the loss of money they are risk seeking. This result is contradicted with assumption in many travel behavior studies that is people evaluate time and money by the same way. Based on the survey data, the weighting function for time gains and time losses is proposed respectively in the paper. We suggest people will use the way of segregate gains to integrate multidimensional gains and losses to form the trade utility, so the functional form of the trade utility is decided. The value of time (Vot)is introduced to convert the gain and loss of time attribute into monetary standard.Finally, time and money budget in the account were treated as constraints, the travel choice behavior model was developed respectively under the definite condition and risk condition. Due to significant non-linear feature of the trade utility function, the Newton-Rapson Algorithm is not suit to estimate the new model unknown parameters. The improved genetic algorithm was used to estimate unknown parameters instead of Newton-Rapson algorithm. Meanwhile, some indexes for evaluating the precision of parameters estimation are given definitely.The Chengdu-Chongqing corridor was taken as study object.Comparison between the new model under the definite condition and the traditional MNL model was made to demonstrate the precision of new model. In order to prove the precision of new model under the risk condition, the commuter's departure time decision was selected as research object, the new model results were compared with the results of Prospect Theory. According to the two kinds of result, we can certify that the travel choice behavior model based on mental accounting is better than the tradition travel choice behavior models which can forecast and describe the travel choice behavior more accuracy.The stated preference is an important way to obtain survey data. The stated preference design is the key factor which can impact the survey data quality significantly. Currently, the study of stated preference design falls behind the research of travel choice behavior model. The commonly used stated preference designs are the random factorial design, orthogonal design and utility balance design. The random factorial design can effectively reduce the number of combinations, but it is unable to avoid the prediction error caused by the correlation among attributes. The orthogonal design can avoid the correlation among attributes, so it is widely used in the study of traffic behavior field. The orthogonal design has a shortcoming that is if too many alternative and alternative attribute levels are set, the number of experimental combinations obtained by the orthogonal design becomes very large, and the survey work would be hard. The random factorial design and the orthogonal design are both introduced from regression analysis which is not suit for the travel choice behavior model. The utility balance design was aimed at obtaining enough trade-off information in data.The trade-off information is the key factor to reflect the feature of travel choice behavior accurately. Because the value of the utility balance cannot be defined precisely, the utility balance design is used fewer. In order to overcome the disadvantages of orthogonal design and to keep enough trade-off information in data, a new stated preference design was proposed. The theoretical basis of the new design is to obtain these combinations when the value of the determinant of the asymptotic-covariance matrix of the model is smallest. The stated preference designs were deduced based on the model structure which can keep the quality of data and improve the precision of model results.The Chengdu-Longquan corridor is selected as research object. The new stated preference design and the orthogonal design were used to design two kinds of stated preference questionnaire respectively and acquire the data. Comparisons between the two kinds of design were made for the model parameter estimations under the large sample and small sample separately. According to empirical results, we found that the new stated preference design can improve the precision of model results, especially under the small sample.In this paper, we developed the travel choice behavior model based on the mental accounting, and the improved genetic algorithm was used to estimate the new model unknown parameters and the empirical study were also completed. The new model overcome the shortcomings of traditional travel choice behavior theory,it can improve the precision of model results from the model basic theory.Meanwhile, the new stated preference design was proposed. The essence of the new stated preference design is to design these combinations based on the travel choice behavior model structure which can improve the precision of model results from the quality of survey data. We improve the accuracy of travel choice behavior analysis results from two aspects: the model basic theory and the survey data in this paper. The research achievements can provide theoretical foundation for making somescientific traffic policies and measures to guide people to travel reasonably.