Multi - Scale Urban Land Use Data Mining

In the era of big data, the departments of land and resources have been built Land resource database of different year, multi-scale and multi-type. However, the studies of these data always stay in the shallow level. How to obtain useful information from the database in-depth, supporting government decision-making of land use, land planning timely and effectively is one of the important land problem to be solved.Based on a large number of experiments, this paper discussed the application of spatial data mining methods in the spatial structure of land used of urban construction, through researches on data scale selection, which kind of mining tools to choose and mining algorithm, combining with the actual case. This paper is divided into three parts:the first part introduced the background and significance of this topic, current situation and problems at home and abroad, and the paper content and structure; the second part focuses on introductions of data mining algorithm in this paper; the third part studied cases, analyzed and mined spatial structure of the urban construction land by Gini coefficient and data mining software Weka. Conclusions obtained are as follows:(1) Analysis of the spatial structure of urban construction land. By using statistical analysis and introducing Gini coefficient in the analysis of the spatial structure of land used for urban construction, Results are got as follows:①National gini-coefficient based on regions is lower than what based on provinces and cities. Gini-coefficient got bigger when dataset is bigger. It indicated that there exited differences in gini-coefficients among different scales which can only be used in single scale.②The land use gini-coefficients based on different scales shows that on nation, region and province these three levels, each land use of urban construction is of uniform distribute. ③On region level, gini-coefficient of central distinct and west distinct got smaller than else, and east distinct got irrational distribution of urban land use.④On city level, most gini-coefficients of cities were decreasing meaning that urban land use distribute rational over country.⑤Compared with the standard of structure of land use of urban construction, results of gini-coefficients proved to be a worthy method to measure whether land use of urban construction distributed rationally.(2)Mining characteristics of spatial structure of urban construction. Conclusion obtained from analysis such as association、cluster、and regression based on Weka: ① According to association analysis, there exists relations among total area of city、 population and land use of urban construction.②According to association analysis, all cities were divided into 5 classes with unique characteristic. In sum, distribution of land use of urban construction is balanced and at the time gathered relatively.③ According to regression analysis, linear relation was found among land uses which can be uses to predict area of certain land use.