基于K-Means聚类的辽宁省主体功能区试划研究

在国家提出的新的主体功能区概念体系和划分思路前提下,基于辽宁省主体功能区划定要求的8个基本指标项定量综合评价基础数据,选择“典型的”优化开发区、重点开发区、农产品主产区和重点生态功能区,采用K-Means聚类方法,分别在不指定初始聚类中心、指定初始聚类中心、以及PCA分析（即主成份分析）基础上指定初始聚类中心3种方法,对于全省100个县市区做4类主体功能区试划分.分别将划分结果与全省最终采用的划分方案进行比较,结果表明采用在PCA分析基础上指定初始聚类中心技术方法的划分结果更接近最终采用方案.

MAJOR FUNCTION ORIENTED ZONING OF LIAONING PROVINCE BASED ON K-MEANS CLUSTERING

The new major function oriented zones concept system and classification method, which proposed by the state is the premise of this paper. Based on eight basic index of Liaoning province, which is satisfy the requirement of major function oriented zoning, it quantitative comprehensive evaluated of basic data and chose the typical optimized development area, key development zones, agricultural products producing and the key ecological function areas. Using K- Means clustering method, it respectively based on without specifying the initial clustering center, specifying the initial clustering center, and the analysis of the PCA （ principal component analysis） to specify the initial clustering center three ways, and according to the four types of major function oriented zones, try to divide unit of 100 counties in Liaoning province. Respectively compared the dividing results with the final classification scheme of Liaoning province, it is clearly that the analysis technology in the basis of PCA to specify the initial clustering center is closer to the final solution.