硕士论文：

作者：杨斌，导师：郑新奇

作者：孙元军，导师：郑新奇

作者：赵璐，导师：郑新奇

作者：李志建，导师：郑新奇

作者：王志国，导师：郑新奇

作者：董金玮，导师：郑新奇

期刊文章：

李志建, 郑新奇, 吕利娜, 等. 基于邻接指数的空间关联规则挖掘方法研究[J]. 测绘科学, 2009, 34(6): 115-117.

【摘要】 在空间关联规则挖掘中一般是采用遍历算法进行,导致对海量数据计算效率的降低。目前,空间数据挖掘模型多采用空间邻接矩阵来表达空间关联权重,大多情况下没有考虑邻接关系的实际量化的结果。文中在分析了空间实体分布的各种相邻关系基础上,采用邻接指数的方式来测算空间相关程度,并在此基础上采用改进的Apriori算法,通过自编程序加以实现。以北京市昌平区土地利用类型的空间分布关系为样例数据进行了试算。结果表明,计算效率有较大提高,并挖掘出一些潜在的土地利用类型间的共生关系。

孙元军, 郑新奇, 常伟倩. 基于 Weka 的城市建设用地结构特征挖掘研究[J]. 计算机工程与应用, 2008, 44(27): 231-235.

【摘要】 城市建设用地结构是城市建设用地状况的直接反映,对其进行研究分析可以了解掌握城市建设用地的现状与内在规律。尝试使用Weka软件,运用其关联分析、聚类、回归方法中的不同算法对城市建设用地结构进行分析,获取了一些有益的结果:(1)通过关联分析得到不同用地间的相关性;(2)通过聚类将全国城市分为5类,并分析了其空间分布特征,即存在均匀分布,又呈现相对聚集现象;(3)通过回归分析得知用地结构与城市规模之间存在一定联系。 更多还原

孙元军, 郑新奇. 基于 SQL Server 的城市地籍空间数据挖掘探讨[J]. 计算机工程与应用, 2007, 43(13): 200-202.

【摘要】 地籍资料记载了土地的基本信息,其数量巨大,使用传统方法不宜处理。而数据挖掘技术可以有效分析大量数据,从中提取有用的信息。论文尝试使用SQL Server2000组件Analysis Manager软件,对地籍信息进行数据挖掘操作。实验结果发现该方法具有一定的实用性,得出的结果具有一定的参考价值。

李新运, 郑新奇. 基于曲边 Voronoi 图的城市吸引范围挖掘方法[J]. 测绘学院学报, 2004, 21(1): 38-41.

【摘要】 在分析城市吸引范围特点的基础上 ,指出了忽视实体属性的普通Voronoi图不能有效地表达城市吸引区空间网络。根据城市引力模型导出了城市吸引范围的边界曲线方程 ,探讨了曲边Voronoi多边形的自动生成方法。以MapInfo为平台 ,通过MapBasic编程计算并绘制了济南市的吸引范围。理论和实践表明 ,位置—属性一体化的曲边Voronoi图可以更好地刻划城市的空间吸引特征。

郑新奇，赵璐.Association rule analysis of spatial data mining based on matlab a case of ancheng township in china.

摘要：Though there are many methods for spatial data mining, it is also a desirable problem that how to carry out this operation fast. Matlab is called full-purpose calculation paper. And it has certain advantage in carrying out data mining because of its powerful matrix calculation function. According to the association rule method of data mining, spatial data of land use and slope was processed and extracted on Geographic Information System (GIS) at first, and data mining was performed by the program, which was wrote in M-language of Matlab. The study area was Ancheng township in Shandong province of China. The confidence and the support between different land use types and slope levels were mined, and the association rules were set up. Through the empirical study, it has been proved that this study method is feasible and the conclusion can provide instruction for local land use planning.

郑新奇，象伟宁，董金玮，A spatial cluster method for prime farmland selection

摘要：In China, the protection of prime farmland is a national policy of eminent importance. From an analytical perspective, its central mission, designating qualified cultivated land as prime farmland, is a comprehensive multi-attribute ranking problem. The paper first analyzed the shortcomings of the existent methods, and then proposed a spatial clustering method for prime farmland designation. The main processes were as follows: (1) Building of index system of prime farmland delimiting; (2) Evaluation of index weights according to expert knowledge; (3) Partition of study area on account of the accuracy. Data of elliptical regions that was preprocessed was put into model of spatial clustering. (4) The land parcels in same cluster is combined into larger units, ranking the result units by holistic productivity level, and selecting the super units which meet the requirement of prime farmland, finally modulating the result according to the correlated policy. The final result can be acquired. Take Jiyang County, Shandong, as a study area, the research showed that the model of selecting cultivated land into prime farmland by GA-K means spatial clustering can effectively solve the existing problems. The prime farmlands were obviously more concentrative, the isolated land parcel fragments were removed effectively, and spatial gathering level was enhanced remarkably. All that makes it a scientifically sound and practically feasible tool to protect and manage prime farmland, and monitor the prime farmland, and realize the aim of scientific and scale management.