基于K-means的最佳聚类数的求解问题研究

针对经典k-means聚类算法的弊端进行一定程度上的改进,提出一种新的基于距离相等函数决定最佳聚类值的改进方法.实验采用两大类标准数据集来测试该算法,并和k-means算法的结果进行了比较,证实了该改进算法的有效性,解决了聚类数目k值的难确定性问题.

A Study on K-means Algorithm for Determining Optimal Number of Clusters

For the improvements of the drawbacks with the classical k-means clustering algorithm to a certain extent,the author has proposed a new and improved method for deciding the same function on the best value of distance-based clustering.By comparing two major categories of standard data sets used to test the algorithm,and k-means algorithm,results have confirmed the validity of the improved algorithm and solve a number of clusters K-means of hard uncertainty.