The Study On Method And Application Of Urban Spatial Data Mining

In recent years, with the establishment of urban database, it appears the situation of "data is abundant, but information is lack". Therefore, it is a new task of geographical information science that adopts the method and technology of DM to support urban development. SDM is attractive and challenging field of study. It has been used in many fields, such as automatic collection of spatial data, dynamic monitoring and time and space prediction, etc. Based on the present study of SDM, this article studies the theory, method and appliance of USMD which includes Urban spatial distribution data mining, Urban spatial dynamic data mining, puts forward some new mining method of SDM, improves the present mining method, carries out a large of practical experiments and researches and fundamentally builds up an experimental system of USDM.The full paper divides into five parts altogether.The first part: Introduction. This part expounds the definition of SDM, concludes the status and trend of SDM in the world, such as the study of SDM theory system, computation of spatial distribution character, spatial classification regularities and clustering rules, spatial association rules and prediction and simulation of spatial structure, discusses the frame structure of SDM and summarizes the basic methods of SDM.The second part: Urban spatial distribution data mining. The task of USDM includes spatial evaluation, spatial classification, computation of spatial distribution, spatial optimization and spatial relational analysis. This part mainly studies the methods and application of spatial evaluation and spatial classification. Adopting the conception model of combination of coordinate and attribute, this part brings spatial coordinates, spatial relationship and attribute character into the unitive model of spatial computation, studies city's land suitableness evaluation and the method of spatial clustering in division of urban function districts and puts forward the arithmetic of classification layer.The third part: Urban spatial dynamic data mining. Spatial dynamic prediction is the main method of spatial dynamic data mining. To prediction and simulation of discrete-state attributes, this part establishes the predictive method of CA which has operation and discusses the method and application of prediction of urban land-use structure taking Jinan as an example. This method resolves automatic collection of spatial state part conversion rule in the present CA that has not matured. To prediction of continuous state attributes, this part puts forward the SAR analytic method of combination of location and attribute and predicts GDP of Shandong's 17 cities using this method. To prediction of city spatial expansion, this part brings forward predicting method of point sources radiation diffusion based on the thought of regional diffusion and takes Jinan as an example.The fourth part: Establishment of urban spatial data mining experimental system. This part initially establishes USMDS by working out the procedures of mining computation and integrating GIS platform and other data analysis software.The fifth part: Conclusion and expectation of study. This part concludes researching production that acquires in this paper. There are many methods in SDM, but this paper only refers to the methods of urban spatial distribution and spatial dynamic data mining. Other methods of SDM, such as spatial association rule,spatial discrete computation and rough set, are further aspect of study in the urban use.