Research On The RP/SP Data Combined Method

The combined use of RP and SP data, which exploits their respective advantages, is an effective method to study new traffic service or policy. Considering the similarity among alternatives in mode choice problems, the Nested Logit is usually applied to construct sub-models. In previous literatures, however, RP and SP sub-models often adopt different hierarchical structures which result in problems such as inconsistent comprehension for behavioral mechanisms, unclear corresponding relation between error-components and incorrectly description of the real hierarchical structure of SP data. In addition, the literature on state-dependence of SP choice is still limited to the inertia effect of actual choice, failing to reveal the influence of internal order-correlation. This paper takes new bus lanes on freeway in Beijing as an example to study the mentioned problems, and the main research contents are as follows:(1) Joint methods using NL model to construct sub-models of RP and SP data along with solutions to deal with state dependence at home and abroad were analyzed and summarized. Modification is proposed based on the limitations of previous study.(2) Questionnaire on commuter mode choice was designed, including RP(Revealed Preference) survey and SP(Stated Preference) survey. Investigating residents within 2km range of new bus lanes on Jingkai freeway to get their present traffic condition, commuter mode and potential demand for rapid transit. The residents’ commute characteristics were analyzed according to the survey data in order to recognize the main factors influencing residents’ commuting mode choice.(3) Based on the results of the survey, RP and SP independent NL model are firstly estimated to identify the true hierarchical structures of both types of data. Then, considering hierarchical scale parameters and fusion coefficients, three different RP-NL/SP-NL models and RP-NL/SP-MNL model are estimated and compared.(4) Traditional method, adding the previous choice dummy variable in the SP model and improved method, considering correlation with error components are applied to reveal the influence of internal order-correlation of SP data.In this paper, the analysis is based on RP/SP combined survey results to construct discrete choice model. The paper hopes to take advantage of RP data to further correct the bias of SP data, clearly reveal the choice process of SP data and improve the credibility of SP data and the accuracy of the prediction of RP/SP joint model.