Research Of Key Techniques In Spatial Data Mining Based On Geographic Information System

Spatial data mining of spatial databases, is the extraction of implicit knowledge, spatial relations and discovery of interesting characteristics and patterns which are not explicitly represented in spatial databases. The technique can play an important role in understanding spatial data and capturing the intrinsic relationships between spatial and non-spatial data. In recent years Geography Information System (GIS) has been used in many fields. It has become one of the important tasks, which need be studied currently, because the amount of spatial data obtained from GIS and other sources has been growing tremendously.This thesis systematically discussed the basic theory of spatial data mining, and proposes a model of spatial data mining based on GIS and several algorithms of spatial data mining from mining relevant spatial knowledge. Base on these algorithms, a spatial data mining tool—GIS_Miner, is suitable for GIS has been developed. The achievements of this dissertation can be concluded as follows.(1) The technology framework of spatial data mining is established and the theory and methods are perfectly developed. The definition and characteristics of SDM are set forth , and the system architecture of SDM system including data source, miner and user interface is proposed. The essential processes and models of SDM are studied and rules are discussed. Many spatial data mining approaches are introduced in this paper and each method's characteristics are analyzed. Some principles of developing SDM system are pointed out.(2) The pattern of spatial outlier detection algorithm based on the attributive correlation is proposed. The outlier detection data and it's mining are defined, the four methods of spatial outlier detection are surveyed. A novel method of Spatial Outlier Detection Algorithm Based attributive correlation is introduced. This approach can effectively solve the drawback. It uses the matrix of attributive correlation and the dynamic index structure which is R-tree on the base of spatial statistics, Some experiments are done on the cadastre data in this method. The range of application and performance of this approach are analyzed, and it can get good results under the multi-attributive correlation.(3) he rule-based spatial co-location method in categorical data is discussed. The spatial association rule methods and co-location algorithms are introduced. The co-location method based on vector data is analyzed. Concerning the categorical characteristics of spatial data. This algorithm defines the transaction of data mining by using spatial relation and finds the co-location rules by using the multi-layer participation index. It can solve the problem effectively. Some experiments have been done on fire data in city in this method. The range of application andperformance of this approach are analyzed, and it is found that the rule-based spatial co-location approach in categorical data is applicable in other continuous data after being discreted, good results can also be obtained.(4) Spatial co-location algorithm based on time-series is proposed. The methods of event sequence and the range are analyzed based on time-series analyzing, time-series spatial co-location method is researched according to the distribution of events and other techniques. Spatial time-series can be effectively analyzed setting the event folding window. The time complexity and experiments are achieved.(5) Based on the research of above algorithms and GIS, a spatial data mining tool-GIS_Miner has been developed using software component. It has a flexible and open architecture, and mine multi-type of knowledge such as association rule, classification rule and time series. The system is used for Hangzhou 119 in china. Results show that it is effective.