For the deficiencies of uncertain data stream clustering algorithm — EMicro in accuracy of clustering results and in probability similarity of data in clustering,we present a new existing level-based uncertain data stream algorithm — UDsStream.By setting the probability threshold,the algorithm separates the processing of the points with high uncertainty and the points with low uncertainty to improve the accuracy of clustering results and the probability similarity of clustering data.At the same time the window and density methods are introduced,by setting the dynamic density threshold the distribution feature of data stream can be dynamically handled,so makes the clustering process have better flexibility.Experimental results show that compared with EMicro,the UDsStream algorithm has better clustering effect.%针对不确定数据流聚类算法———EMicro在聚类结果的精确性和聚类中数据的概率相似度方面的不足，提出一种新基于存在级的不确定数据流算法———UDsStream。该算法通过设置概率阈值，将不确定性高的点和不确定性低的点分开处理，提高聚类结果的准确性和聚类中数据概率相似度。同时引入窗口和密度的方法，设置动态密度阈值，能够动态掌握数据流的分布特征，使聚类过程有更好的灵活性。实验结果证明，与EMicro相比，UDsStream算法聚类效果更好。
展开▼
