Determining the Best Clustering Number of K-Means Based on Bootstrap Sampling

In this paper, the selection of clustering number in K-means clustering algorithm is studied, based on the Bootstrap sampling, a new method is proposed to determine the best clustering number based on the between the actual value of total within-cluster sum of squares and its estimated interval. By UCI Machine Learning Repository and randomly generated artificial simulated test data sets, the experimental results show that using clustering has obvious improvement, the method can overcome falls into local optimum caused by unreasonable clustering number to select. K-means algorithm.