基于聚类分析的土地社会功能分类研究

Classification of Land Social Functions Based on Clustering Analysis

为提高土地社会功能分类的科学性和合理性,探索新的土地社会功能分类方法,为土地社会功能分类研究提供技术支撑。通过文献共收集22项土地社会功能,针对各项功能自身具有的不同属性,提出了10项土地社会功能特征因子。对各项土地社会功能的特征因子进行赋值从而获得土地社会功能评分表,继而进行谱系聚类和K均值聚类分析。运用数理分析中常用的R^2、半偏R^2、伪F和伪t24个统计量对谱系聚类结果进行最佳聚类项数分析,采用R^2和伪F两个统计量对K均值聚类结果进行最佳聚类项数分析,得出3种谱系聚类结果和1种K均值聚类结果与顺序法分类结果比较分析,最终提出将22项土地社会功能分为8类的分类结果。研究表明聚类分析可很好地运用于土地社会功能分类研究中。

In order to improve the classification and rationality of land social function,this paper explores the new classification method for land social function,and provides technical support for the research of land social function classification.A total of 22 land social functions were collected by reading literature,and 10 land function characteristics were put forward according to the attributes of each function.The land function scoring tables were obtained by assigning the characteristic factors of each land function,followed by phylogenetic analysis and K-means clustering analysis.By using the statistics of R^2,semipartial R^2,pseudo-F and pseudo-t2,which are commonly used in mathematical analysis,the best clustering analysis was made on the pedigree clustering results.Using R^2 and pseudo-F statistics,the result of the cluster analysis of the three kinds of pedigree,one K-means clustering result and the sequence method were compared and analyzed,and finally 22 kinds of land social functions were classified into 8 categories.The results show that clustering analysis can be applied to land social function classification.