K-Means Clustering With Incomplete Data

Clustering has been intensively studied in machine learning and data mining communities. Although demonstrating promising performance in various applications, most of the existing clustering algorithms cannot efficiently handle clustering tasks with incomplete features which is common in practical applications. To address this issue, we propose a novel K-means based clustering algorithm which unifies the clustering and imputation into one single objective function. It makes these two processes be negotiable with each other to achieve optimality. Furthermore, we design an alternate optimization algorithm to solve the resultant optimization problem and theoretically prove its convergence. The comprehensive experimental study has been conducted on nine UCI benchmark datasets and real-world applications to evaluate the performance of the proposed algorithm, and the experimental results have clearly demonstrated the effectiveness of our algorithm which outperforms several commonly-used methods for incomplete data clustering.