基于算法选择和结果评估的自动聚类方法

摘要

An automatic clustering method based on clustering algorithm selection and result assessment is proposed. The method first analyzes the potential cluster structure of the given data set,and selects an appropriate set of alternative clustering algorithm for the data set based on the found cluster structure. Then it uses the cluster validity index to assess the clustering results of the clustering algorithms to ensure high-quality clustering results. The experiment results show that the proposed method can select the clustering algorithm successfully, and get high quality clustering results.%文章提出了一种基于算法选择和结果评估的自动聚类方法.对给定数据集,该方法首先通过分析数据集的潜在簇结构,并依据所发现的簇结构为数据集挑选一种合适的备选聚类算法集；然后利用聚类有效性指标对这个算法集的算法聚类结果进行评估,以确保得到高质量聚类结果.实验结果表明该方法能够自动地挑选适合数据集的聚类算法,并获得高质量的聚类结果.