The Development And Application Of Spatial Data Mining Prototype System

Either Data Mining ,spatial data mining or visualization of geographic information, the purpose is to promote the scientific understanding of spatial data and construct knowledge system in high level. It is a process of Step-by-step and continuous optimizing. Although the theory of spatial data mining has been developed for many years, we have to rely on the development of computer science only if it wants to service for daily life. With the development of Computer software and hardware technology, mining algorithms tends to be more and more open. Reduction of software developed cost also makes professional-oriented spatial data mining system, with each passing day. In view of this, this paper is funded by the National Natural Science Foundation Project (40571119), developed a data mining software within spatial planning field base on existing map engine, which is provided technical support and solutions to decision-making management.To begin with, this paper analyse and summarize a large number of documents, present the status and development trends of spatial data mining at home and abroad, after a grasp on the overall direction and existing spatial data mining software features, which was designd and realized the spatial data mining system. The main contents includes: Forecasting spatial data through classification under conditions of Non-linear. It is a precise algorithms of neural-net to forecast spatial trends, because the algorithm has more space complexity, require for higher computer hardware quality. Now Exchanging space for time is a kind of design pattern of software. Followed by association-rule mining that based on self-adjacent, Self-adjacent index is a kind of Coefficient based on autocorrelation in Nominal Scale. Taking self-adjacent as object of association-rule mining, we find out the Symbiotic relationship between Land-use types. It is provided a kind of thought for the adjacent index and combination of association rule mining. We utilized spatial data visualization technology in order to display the results of clustering, Grid feature and Gini coefficient,which were research parts of this paper.In last part of the paper, the application of five specific cases illustrate the practical significance and the value of this article First of all, the utilized of high spatial complexity of neural network algorithm on a relatively small amount of data have a good accuracy and efficiency. Mining results also reflect the value of new ideas that combine autocorrelation and association rules, in terms of running time or the type of symbiosis results has proved the feasibility of the method. The visualization simplified operation of spatial data mining process for users.