Research On Discovery Of The Problems Of Land-use Based On Data Mining

With the rapid development of economy and the urban population,the urban spatial expansion is a kind of inevitable trend.While occupation and waste of urban land occurred frequently,which are faced to the government.The data of land-use is the important fundamental of land resource management.A wide range of information is covered in the data throughout the business process of "batch of supplies for review".So research on the discovery and prediction of problem is quite significant to the effective use of land resourcesIn this paper,data mining is applied to discovery and prediction of problematic land based on the land database of Hannan District in Wuhan,and different solutions for different types of problematic land is put forward.The main contents of this paper are as follows:(1)The land-use problems typically occurred in the land business are summarized,and the types of land-use problems are divided from the point of view of data mining.Different research methods for different types of land-use problems is put forward.(2)An expert system based on rule judgment for the discovery of land-use problems is designed.In the expert system,the knowledge and experience of experts are collected to establish the library of rules.The design and implementation of the mode for discovery based on experts'experience and rules is described in the paper,including the functions:add and modification of rules,add and modification of data of land,the report of discovery of land-use problems and other functions.(3)Taking idle land as an example of the predicted type of land use problems,Random forest,gradient boosting decision tree(GBDT),support vector machine(SVM)is estimated for the discovery of idle land in this paper,and the relationship between the size of the attribute space and the classification of idle land is analyzed.In addition,an algorithm based on multi-classifier fusion with soft voting way is proposed to improve the accuracy of prediction idle land.(4)On the basis of data mining,expert experience and data mining are combined,the knowledge of land evaluation is applied,and accuracy of the prediction of idle land by data mining algorithm is optimized,meanwhile,the performance of the experts' knowledge and data mining fusion algorithm is also demonstrated.For different types of land problems,the corresponding methods is put forward in this paper.And the anticipated goal is achieved,find the problems in land management land provides a scientific and effective help.