Abstract:

The initial clustering center of the traditional K-means algorithm is generated randomly from the data set,and the clustering results fluctuate with the initial cluster centering of different.A new improved K-means algorithm based on density is proposed,by which the density parameter and neighborhood distance of every data object is computed,then k point in high density parameter are chosen as the initial clustering centers.A comparison is made using UCI database as testing datasets with given number of clusters.The clustering results demonstrate that the improved algorithm can enhance the clustering stability and accuracy relatively.

Key words: K means algorithm;clustering center;density parameter;neighborhood distance