Efficient Handling of Incomplete basic Partitions by Spectral Greedy K-Means Consensus Clustering

Cluster ensemble approaches are combining different clustering results into single partitions. To enhance the quality of single partitions, this paper examines a comparative study of different methods with advantage and drawbacks. Performing spectral ensemble cluster (SEC) via weighted k-means are not efficient to handle incomplete basic partitions and big data problems. To overcome the problems in SEC, Greedy k-means consensus clustering is combined with SEC. By solving the above challenges, named spectral greedy k-means consensus clustering (SGKCC) is proposed. The proposed SGKCC efficient to handle incomplete basic partitions in big data which enhance the quality of single partition. Extensive evaluation NMI and RI used to calculate the performance efficiency compared with existing approach proving the result of proposed algorithm.