Research on Sampling Method of CFSFDP Clustering Algorithm and Its Criteria for Determining the Best Sample Size

Abstract

Clustering by fast search and find of density peaks (CFSFDP) is a novel density-based fast clustering method, which has been widely studied and applied in many fields. However, when the sample size of data is too large, the algorithm is inefficient, since it consumes a lot of time and storage space. To solve the above problem, a simple random sampling (SRS) method is provided to speed up the optimized CFSFDP algorithm for real data with large sample size. The rate of correct classification of the sample is defined to measure its clustering performance, and we call it as sampling accuracy. We first use SRS method to generate small samples for cluster analysis. Then, we explore the relationship between the sampling rate and the sampling accuracy. Finally, in order to determine the best sample size that can achieve high sampling accuracy with high efficiency, the mean and standard deviation of the sampling accuracy are adopted as two criteria, and the best sample size is determined based on them. A real case study is given to show the implementation and effectiveness of the proposed method.