基于特征权重优化的K-中心点聚类算法

It gives a kind of K-medoids clustering algorithm based on feature weight optimization(KMCAFW), described by algorithm 1 and algorithm 1*, in order to find more meaningful clustering distributions by searching a better dissimilarity measure in hybrid attributes data space. After some necessary discuss, it lists its time complexity and astringency analysis. In order to implement the feature weighting part in this new clustering algorithm, it gives two different methods optimizing feature weight, described by algorithm 2 and algorithm 3 each. It validates the new clustering algorithm can often get a better clustering quality by two control experiments using Iris and German data sets. At last, it indicates several valuable research expectations.