Departmental Bulletin Paper A Ranking Fuzzy Cluster Algorithm for Multidimensional Data

Guo, Jia

Abstract— Data classification is an important problem in various scientific fields. Data analysis algorithm such as the fuzzy c-means algorithm and k-means algorithm are wildly used on multidimensional data but still suffers from several drawbacks. To get higher classification accuracy, the Ranking Fuzzy Cluster (RFC) algorithm which uses the method Dimension Ranking (DR) and a new center moving formula is proposed in this paper. The method DR is used to calculate the weight of each instance among all the data set in each dimension. The new center moving formula which uses the non-gradient descent iterative to avoid the local minimum problems of FCM. In the experiment, five benchmark data sets are used to test the classification accuracy of RFC. The fuzzy c-means algorithm and k-means algorithm are also used in the experiment comparison. Experimental result shows that the RFC has higher classification accuracy than fuzzy c-means and k-means with multidimensional data. Keywords—Fuzzy Cluster; Multidimensional Data; Dimension Ranking; Cluster Center Moving;

Number of accesses :  

Other information