Research On Spatio-temporal Meteorological Date Mining Technology Based On DSCAN Optimization Algorithm And Decision Tree Optimization Algorithm

DBSCAN and Decision Tree are two important technologies in spatiotemporal data mining.In application of data mining technologies,valuable information can be analyzed,generalized and revealed from a great amount of intricate and seemingly unrelated data.GIS has great power of storage,management and analysis in handling temporal and spatial data.Compared to traditional data mining technologies which only work on meteorological factors,Space-time meteorological mining of GIS can analyze and apply temporal,spatial and attributive factors of meteorological data.A combination of data mining and GIS not only can reveal the underlying rules concerned with both temporal and spatial factors,but also offers new technology and wide application.However,spatial-temporal clustering algorithm and association rule mining have not made full use of the spatial analysis capability of GIS,and they are seriously redundant in processing,thus their computing power can still be improved.Aimed to improve the low efficiency of previously mentioned technology,this thesis makes a study on the meteorological factors in southwestern China,and proposes a method combining temporal-spatial association rule and temporal-spatial clustering rule with Grid image processing technology and Vector data processing technology which has been applied to an intensive analysis of ten thousands sets of data collected from the areas of Yunnan,Guizhou,Sichuan and Chongqing provinces.It has been found that the processing increases the utilization of historical meteorological data and makes more use of GIS in weather forecasts and climate researches.Therefore,this thesis focuses on solving the problem of spatiotemporal association rule mining,spatio-temporal clustering rule mining,key technologies and algorithms for meteorological spatio-temporal data,and makes the following researches:Making a summary of previous research accomplishments and pointing out their limitations,which is based on a literature review of temporal-spatial data mining technology combined with GIS,and its application in processing meteorological data;Based on the classical clustering algorithms,this thesis puts forward a new algorithmintegrating classical DBSCANclustering algorithm,extended Raster searching method and DBSCAN scanning algorithm,which manages efficiently the partition of temperature factors in time-space dimension and their union.It has been proved to be efficient and practical.After analyzing and comparing with the previous decision tree algorithms,this thesis proposes Topology decision tree algorithm which combines the classical C4.5decision tree association rule mining algorithms and theSpatial topological relation query.It gives an efficient solution to mining problems of Mining problem of spatio-temporal partition Association and proves to be efficient and practical.On the basis of ArcEngine by adopting C# programming language,this thesis adopts SOL Server ArcSDE to manage spacial database,and finishes the preprocessing of meteorological data,space interpolation and rule mining.Meanwhile,the thesis also develops essential program plugin which help to reserve and visualize relevant knowledge.This thesis has strong theoretical and practical value to meteorologicaltemporal-spatial data mining based on GIS spatial statistics.This thesis also serves as a modest spur to further studies in temporalspatial laws of meteorological factors in southwestern China.To some extent,it enriches and perfects the theoretical system and technology framework of data mining...