Data Mining Method Based On Spatial Database Research

With the fast development of spatial data obtained technology,spatial data increase up rapidly.Spatail Data Mining technology was mentioned In order to use the resources in the spatail databases enoughly. Spatail Data Mining technology could help people understanding spatial data and picking up relations between spatial data.In ths paper,we introduce the basic theory of spatial database and spatial data mining,include the spatial data structrue,spatial data model,spatial data index technology and the steps ,methods of spatial data mining.All of these are the basic theory of cluster algorithm.Clustering analysis is a very important field in spatial data mining area.It depends on measure the similarity between spatial datas into meaningful subclasses so that the members of a cluster are as similar as possible whereas the members of different clusters differ as much as possible.Spatial clustering is widely used in daily life.it can be used in location selection,customer classification and so on .It can help investors'decision-making, and bring benefit as much as possible. In synthesis,clustering has important Research significance.Now,there are many mature clustering algorithms,such as DBSCAN algorithm,CURE algorithm,CLARANS algorithm and so on.These are classic algorithms in clustering filed,but there are still many Challenges.The most important reserach in this paper is to improve the efficiency of clustering algorithms.In this paper,we introduce two improved algorithms for the general clustering and for the clustering with obstacle constraints.Algorithm 1:impove the DBSCAN algorithm.DBSCAN algorithm repeatedly picks every points and examining whether it is a core point,the I/O spending of this step is very big and the step limits the algorithm efficiency.The impoved algorithm needn't to judge whether each of the point is the core point. In the search for regional connectivity,each cycle select a point without a mark:if the point is a core point,and its EPS-region has other points which has other marks,then all the points mark the minimum;if not,select the next point to charge.The algorithm not only reduce the amount of the point which need to judge,but also improve the efficiency.Algorithm 2:clustering algorithm based on Mathematical Morphology in the presence of obstacles.The algorithm based on the idea of the algotithm of MMC,and adding a constraint to deal with obstacles.The algotithm is different from the DBCluC,it is need not to connect each two points to judge whether the points are in a same class or separated by obstacles.It used a Structural element,deal with points which are affected by the obstacles only.If the structural element is crossed with one obstacle,then connect the center point and other points in the circle,if one line is crossed with the obstacle ,then take its flag as false,it's said that the point is in the other side of the obstacle;if not,the point is in the same class with the center point. After analysis, the efficiency of algorithm is superior to other algorithms.In the end,we take a data experiment, verified the correctness and effectiveness of Algorithm .In this paper,We discuss spatial database,spatial database mining,spatial database clustering step by step,and then put forward an improved clustering algorithm. In the future research work,Ineed to read a large number of clustering technology books and articles, Made faster, easier understanding algorithms,and used in real production, life, to support decision-makers to obtain better results.