基于密度指标的大样本数据集聚类方法

摘要

为提高大样本数据集的聚类准确率，减少获得高精度聚类中心的计算量，构建基于密度指标的大样本数据集聚类方法。给出高斯密度函数对加权系数的聚类中心确定方法，采用基于密度指标的加权聚类算法进行聚类，获得准确度较高的大样本数据集的聚类中心。选用标准的 UCI多组数据集对该聚类方法进行验证，与其它文献方法做实验对比。实验结果表明，采用该方法能获得较高精度的聚类中心，提高大样本数据集分类的准确率。%To improve the clustering accuracy of large sample data sets,and reduce the computational burden of obtaining a clus-tering center with high precision,a clustering method of large sample data sets based on density index was built.The method for Gaussian density function to determine the weighting coefficient was introduced,according to the actual distribution features of a data set in reality.And weighted clustering algorithm based on density index was adopted to obtain a clustering center with rela-tively high accuracy compared with other literature methods by experiment in which the standard UCI multi-group data set was used.Experimental results show that the designed method can obtain a clustering center with relatively high precision and im-prove the accuracy of classifying large sample data sets.