Abstract: Aiming at the problems of data sparsity and high dimension, the author proposed a mixed matrix object data clustering method based on k-multiple  value  representation, which effectively  reflected the distribution of cluster centers and matrix objects in clusters. In order to fully mine the hidden information between data, firstly, the algorithm defined the dissimilarity measure between two numerical matrix objects, and gave a heuristic method to update the clustering center. Secondly,  a mixed matrix object data clustering algorithm based on k-multiple value representation was proposed. Finally, the simulation experiments were carried out on real data sets and composite data sets. The experimental results  show that the proposed algorithm can effectively realize data clustering including a large number of records, matrix objects and attributes.

Key words: multiple value representation, clustering, sparsity, matrix object