文档介绍：,以及同一个类别内部的一致性,并进行计算,,在分析层次聚类算法和神经网络的ART2算法的基础上,,,这里结合人类视觉感知理论,提出了获取最优聚类的条件,从而实现了最佳的分类.
本文的主要研究工作如下:
第一章:简单介绍了聚类分析基本思想,研究的内容,应用的前景,发展现状,并且细致的说明了聚类分析的分类统计量——距离与相似系数.
第二章:说明了层次聚类分析的定义及研究方法,对层次聚类分析方法的有效性做出了细致的研究,并提出了基于相似矩阵的有效性函数.
第三章:在第二章层次聚类分析的有效性研究的基础上,提出了一种利用神经网络的方法对数据进行预处理,从而得出初始聚类结果的改进的层次聚类分析方法.
第四章:在前三章的基础上,将层次聚类分析方法应用在电价区域的空间尺度划分问题中,进而实现了电价区域的划分.
关键词层次聚类分析;有效性;神经网络;空间尺度
Abstract
By hierarchical clustering analysis algorithm, the similarity between samples by the definitions of sexual relations after the separation between class and class, and the same type of internal consistency, which makes the calculation process is simplified. In order to achieve clustering, hierarchical clustering analysis and work ased on ART2 algorithm a modified hierarchical clustering algorithm. Price based on the practical problems of regional division, where bination of human visual perception theory, the conditions for obtaining the optimal clustering,in order to achieve the best classification.
The main research work are as follows:
Chapter I: a brief introduction the basic idea of cluster analysis to study the content, applications, prospects, development status, and detailed description of the cluster analysis classification of statistics the distance and the similarity coefficient.
Chapter II: Definition of hierarchical clustering analysis and research methods, hierarchical clustering analysis on the effectiveness of research and made a similar matrix based on the validity of function.
Chapter III: In the second chapter the effectiveness of hierarchical clustering analysis based on the study of a method using work to preprocess the data obtained on the initial clustering results modified hierarchical clustering analysis.
Chapter IV: In the first three chapters, based on the hierarchical cluster analysis method i