摘要
针对K-means算法易受随机选择的初始聚类中心的影响和划分准确率不高的缺点,给出了一种改进的K-means算法。首先对初始聚类中心的选择过程进行了改进,然后对各样本点间差异最大的维进行加权处理。在Iris数据集上对原始算法和改进后的K-means算法的聚类结果进行对比分析。实验证明:改进后的算法稳定,且聚类的准确率达到了92%。
K-means algorithm is a commonly used clustering algorithm based on partition, for k-means algorithm is vulnerable to random selection of initial cluster centers and the division accuracy is not high, so this paper gives an improved k -means algorithm. First, the choice of initial cluster centers has improved, and the various dimensions of the sample points were weighted. The results on Iris data set used the original algorithm and improved algorithm for k-means clustering are compared and analyzed, and experiments show that the improved clustering algorithm is stable and the accuracy rate is 92%.
出处
《重庆理工大学学报(自然科学)》
CAS
2013年第4期77-80,共4页
Journal of Chongqing University of Technology:Natural Science