基于PCA+GRNN+SOFM的入境旅游空间分布研究

The inbound tourism spatial distribution based on PCA+GRNN+SOFM

选取2011年中国31个省级行政区的入境旅游数据,先采用GRNN,SOFM方法对原始数据进行系统聚类,初步分析其空间分布特征,并基于GRNN网络得出合适步长,在此基础上采用主成分分析(PCA)方法实现变量关系的正交,以排除变量间共线性对分析结果的干扰,然后结合GRNN,SOFM方法对数据进行系统聚类.结果表明,采用PCA+GRNN+SOFM的方法可以较好地解释中国入境旅游的空间分布格局,并且空间分布呈现明显的"人"字形格局特征.

This paper selected the 2011 inbound tourism data of 31 provincial level administrative regions in China to analyse the spatial distribution characteristics of inbound tourism through the GRNN,SOFM and hierarchical clustering method.Firstly,the suitable step length from the GRNN network is fixed,and then the principal component analysis（PCA）method is used to achieve orthogonal variable so that overcome the collinearity between the variables.In addition,the GRNN,SOFM and hierarchical clustering methods are applied to study the spatial pattern of Chinese 31 provinces.The results show that using PCA,GRNN and SOFM methods can explain the spatial distribution pattern of Chinese inbound tourism better.And the inbound tourism present obvious ＂human＂glyph structure characteristics.