Research of Similarity Measurements in the Clustering Analysis

Abstract

Similarity measurements play an important role in the clustering analysis, so any good or bad methods of measuring similar degree directly affect the clustering algorithm. In the paper, several approaches to similarity measurements for single attribute type data, which had been proposed, have been discussed. Moreover, a way has been obtained so as to calculate the similar degree of multiple attribute type data. At last a experiment was tested. The result shows that the method is not only feasible but also effective.