基于时间序列的AP-NN混合模型聚类

仿射传播算法是一种快速有效的聚类方法,但其聚类结果的不稳定性影响了聚类性能。对此,提出基于近邻的仿射传播算法(AP-NN),通过仿射传播算法产生初始簇,并从中选择代表簇对非代表簇的样本进行近邻聚类。在时间序列数据集上的实验结果表明,AP-NN模型算法能够产生较好的聚类结果,适用于聚类分析。

Clustering of combining AP with NN model based on time series

Affinity propagation algorithm is a fast and efficient clustering method. However, the stability of clustering results affect its performance of clustering. Thus, a new method combining affinity propagation and nearest neighbor is proposed, this algorithm produces initial clusters through affinity propagation, and then the samples of non-representative cluster clusters to representatives cluster through nearest neighbor clustering. In the time series data set, experimental result shows that AP-NN algorithm can produce effective clustering results and is applied to cluster analysis.