Residential Location Model Based On Individual Decision Behavior: Formulation And Application

The rapid growth of urban economy, accelerated development ofurbanization and extension of urban scale in China are accompanied by anequally rapid growth of motorization. At present, the urbanization of most bigcities in China is experiencing the spatial centralization and decentralizationof urban land-use pattern, whose evolving direction is not very clear in thisperiod. This has made the problem of how to harmonize the development ofurban land-use and transport system an urgent and hot spot of today. How toquantitatively depict the inherently complex and unpredictable nature of sucha dynamic relationship has always been the focus and difficult point in thisfield. From the new perspective of individual decision behavior, this paperproposes an urban residential location model with its algorithm andapplication under the framework of urban land-use and transport integratedplanning concept model.In this thesis, existing discrete choice model literature is first reviewedby classifying the various approaches into two groups: models based ongumbel distribution hypothesis and normal distribution hypothesis. Classicmodels and development of each group is discussed in-depth. The keyobstacles of the first group of discrete choice models are then summarized.Moreover, the thesis focuses on the key obstacles of discrete choicemodel and brings forward corresponding solvable tactics. A new discretechoice probability formulation with basic principle of GEV family modelembedded is presented to study the individual family behavior on the urbanresidential location choice, then, two kinds of improved models consideringthe diversity of individual characteristics and activity schedule of each familymember, respectively. Insights into the models are obtained by investigatingits mathematics nature and studying the main factors that affect the residentiallocation pattern.To estimate the parameters of the above model, a maximumsimulation-assisted estimation method is explored based on computerizedrandom draws and the core of the algorithm such as the procedures for takingquasi-random sequences that can provide greater simulation accuracy andquasi-newton numerical method are discussed in-depth. According to theexample of Dalian city, this paper realizes a program written in Visual C++combined with Access database in order to make the estimation proceduresimple and convenient, the whole procedure of residential location choiceincluding specification and estimation is implemented on the case study ofDalian city to obtain issues necessary for the model implementation anddemonstrate the applicability of this procedure on real problems.Finally, this paper studies domestic and international urban land-use andtransport planning modeling theory and experiences, and then proposes aframework of integrated planning concept model in which the residentiallocation model is incorporated. The author expects that the development offuture urban land-use and transport system integration model and its keytechnologies can benefit from the useful conclusion made in this paper.