关键词: 分类数据, 数据挖掘, 聚类, 不确定数据

Abstract:

Aimed at processing the uncertain categorical data, an efficient uncertain data clustering algorithm, the USqueezer algorithm, was proposed based on the squeezer algorithm. First, this algorithm computed the sum of similarity probability between  uncertain categorical data and each existing cluster. Comparing the largest similarity with a given threshold, it was found that if the largest similarity was greater than the threshold value, the uncertain data would be assigned to this cluster, otherwise the uncertain categorical data was created as a new cluster. Experimental results showed that this algorithm could be effectively used in clustering the uncertain categorical data with a small amount of memory and time.

Key words: uncertain data, categorical data, data mining, clustering