A Dynamic Clustering Algorithm Based on Small Data Set
Tao Peng, Minghua Jiang, Ming Hu · 2009
The traditional clustering algorithms are designed for large dataset or vary large dataset. It is not easy to cluster the small dataset because of the loss of the statistical character and probability character. In this paper, the class ration is introduced, based on the class ratio, the dynamic clustering algorithm is proposed. The dataset are divided into all possible classes, and the class ratios are computed, the min class ratio is chosen and the clustering about the min class ratio is the best clustering. With the experiments, the schema is an effective way for the clustering of small data sets.