An Adaptive K-means Clustering Method Based on Distance Threshold
An adaptive K-means clustering approach based on distance threshold was proposed to get a proper clustering result. Firstly, a reasonable distance threshold was obtained from a given dataset. Then the initial clustering centers were set based on the distance threshold. Clusters with centers close to each other were merged to a new clustering center. Experimental results proved that the suitable value of k and clustering centers could be found with the proposed method, and outliers could effectively be avoided be-ing clustered incorrectly,thus the cluster effect was improved.
国家科技期刊平台
NSTL
万方数据
维普
