基于PCA和决策树模型的农用地(耕地)质量评价研究

主要利用ArcGIS、PCA和See 5决策树模型,以原始数据为基础,对福建省将乐县农用地(耕地)质量进行评价.首先根据将乐县的自然、社会和经济等各方面情况,初选排溉条件等9个主要指标作为评价因子;然后用ArcGIS矢量化提取相关图层并进行相应的空间分析,经过多图层叠加确定评价单元及其属性数据;再利用PCA方法最后确定耕地质量评价综合因子;接着利用数据挖掘工具See 5,自动构建决策树模型,进而用构建的模型来获取耕地质量分等规则的方法;最后划分耕地质量级别,并进行抽样检验结果的准确性.研究结果表明:该县耕地质量中等,一至四等地分别占22%,44%,15%,19%,经实地检验评价结果基本与实际情况相符.

Research on Cultivated Land Quality Evaluation Based on PCA and Decision Tree Model

The thesis involves a tentative research into the cultivated land quality evaluation with ArcGIS,PCA and See 5 decision tree model,on the basis of the original data,and Fujian's Jianle County is taken as an object of study.First,nine major evaluation index-irrigation,height,slope,organism,types of physiognomy,types of soil's character,thickness of tilth,distance to town and distance to highway,for example,are select by Jianle County's environment,community and economy.Second,extract correlative graphs with the aid of ArcGIS vectorization,put them under spatial analysis and use the multi-map superimposition method to define evaluation units and extract unit attribute data.Third,Use PCA method to determine integrated factors affecting cultivated land quality.Fourthly,use See 5 to construct automatically the decision tree and obtain the technique of classifying cultivated land quality.Finally partition levels of cultivated land quality and many samples were collected randomly to verify veracity of the results.The researchful results indicate that the cultivated land in the research region is medium in quality and can sorted into four levels,accounting for 22%,44%,15% and 19%,respectively,which is roughly in accord with the actual situation.