Urban Service Areas Classification And Accessibility Research Based On Data Mining

With the continuous development of the modern city, the demand for the high quality of services has been increasing rapidly. Researches in city planning and resources allocation for the city are becoming more and more active. Taking the financial service system in Chengdu city as an example, we describe how to use the Poisson process to achieve the division of financial service in this paper. On the basis of the relationship between population density and spatial distribution of the financial service system, we propose the logistic regression model for spatial predicting. Also we analyze the accessibility of service issues, basing upon the supply and demand for the financial service. The main contents of the paper consists of three important parts.Firstly, with the characteristics of the local distribution of financial services, we analyze the complete spatial random pattern and the spatial aggregation pattern, proving that homogeneous Poisson process is an effective representation to the completely spatial stochastic model. On this basis, non-homogeneous Poisson process is presented to indicate the aggregation of financial services, and the strength function is used for city service division. Maximum likelihood estimation and kernel smoothing estimation are presented to estimate the strength function in the paper.Secondly, in order to elucidating the factors that influence the result of financial services division, we propose the hypothesis that there is a certain correlation between the financial service points and the distribution of the population. On this base, we build a logistic regression model to analyze the potential relationship between population density and financial service point patterns. Also an effective method is proposed to overcome the defect of the logistic regression model.Thirdly, basing on the Huff model, we define the accessibility of financial services as an indicator to measure the service power that a community can gain. On the premise of simplified model, we analyze the service power that the communities have obtained in Chengdu, and by comparing different models, we realize the purpose for discovering different places which have better accessibility of financial services.Since there is a certain correlation between the urban service systems and the distribution of the population, the spatial distribution patterns are so alike that the models we’ve proposed in this paper are universal. By analyzing and mining the patterns in city service systems, we can also develop the issues about city planning and allocation of resources deeply and effectively.