Research on selecting initial points for k-means clustering

Clustering analysis is one of the important problems in the fields of data mining and machine learning. There are many different clustering methods. Among them, k-means clustering is one of the most popular schemes owing to its simple and practicality. This paper investigates the approximate algorithm for the k-means clustering by means of selecting the k initial points from the input point set. An expected 2-approximation algorithm is presented in this paper. Meanwhile, an efficient algorithm for selecting the initial points is also proposed. At last some experimental results are given to test the valid of these algorithms.