基于互近邻相对距离的最小生成树聚类算法

摘要

针对互近邻距离的不足,提出了互近邻相对距离的概念,同时设计实现了一种新的最小生成树聚类算法.针对某些数据的不平衡问题,提出了兼容不平衡数据的最小生成树分割方法.算法设计简单,易于实现.实验结果表明,该算法能够聚类任意形状数据和兼容处理不均衡数据.对于具有良好几何形状的数据,该算法能够达到非常好的聚类效果,总体性能优于其他算法.%For the lack of mutual neighbor distance,the concept of mutual neighbor relative distance was proposed.To the problem of imbalanced data,a minimum spanning tree clustering method was proposed.This algorithm was simple and easy to implement.The experimental results showed that the algorithm could cluster arbitrary shape data,and deal with imbalanced data.For the data with good geometry,the algorithm could achieve very good clustering effect,and the overall performance was better than other algorithms.