提出了一种利用重心优化初始聚类中心的算法BKM(Barycenter K-Means)。首先将每个候选点临域内所有数据点的重心作为初始聚类中心,然后引入MapReduce进行并行处理计算。结果表明,BKM算法选取的初始聚类中心更为合理,取得了较好的聚类效果。
A Barycenter K-Means(BKM)algorithm with optimized center as initial clustering center is proposed in the paper.At first,the center of all data in the selected region are taken as initial centers and then the parallel MapReduce Processing Framework is introduced into the algorithm.The experimental results show that the selected initial clustering center of the BKM algorithm is reasonable and better clustering results are obtained.