Parallel k means Clustering Algorithm on SMP

The k-means clustering algorithm is one of the popular and simplest clustering algorithms. Due to its simplicity, it is widely used in many applications. Although k-means has low computational time and space complexity, increasing the dataset size results in increasing the computational time proportionally. One of the most prominent solutions to deal with this problem is the parallel processing. In this paper, we aim to design and implement a parallel k-means clustering algorithm on shared memory multiprocessors using parallel java library. The performance of the parallel algorithm is evaluated in terms of speedup, efficiency and scalability. Accuracy and quality of clustering results are also measured. Furthermore, this paper presents analytical results for the parallel program performance metrics.