基于层次分析法的加权聚类融合

WeightedCluster Ensemble Based on AHP

聚类是发现数据分布和隐含模式的一项重要技术,但单一的聚类算法很难达到预期的效果.基于四种聚类算法,根据算法准确率,利用层次分析法以估计权重,构造一个基于四类算法融合的聚类器,经实验检验,该聚类器比单一聚类器的聚类效果好.

Clustering is a technique for the discovery of data distribution and cryptic data pattern, yet clustering algo- rithm alone is not always an ideal solution to achieve the expected result in practice. Based on four clustering algorithms, and the accuracy rate, a clustering aggregation integrated with the four algorithms is constructed via estimating weight of analytic hierarchy process, thus a better clustering result is achieved through experiments.