Call for Papers Editorial Board Content List Subscribe Submission Paper Status Paper Review
A Hybrid Clustering/Classification Method for Complex Datasets Based on Variable Precision Rough Sets and Validity Index Function
Author： Chuen-Jiuan Jane
Keyword： Particle swarm optimization， Variable precision rough sets， PSOVPRS index method， Classification， Clustering
Summary： This paper introduces a hybrid clustering/classification method for successfully solving thepopular clustering/classifying complex datasetsproblems. The proposed method for the solution of theclustering /classifying problem, designated asPFV-index method, combines a particle swarmoptimization (PSO) algorithm, Fuzzy C-Means (FCM)method, Variable Precision Rough Sets (VPRS) theoryand a new cluster validity index function. This methodcould cluster the values of the individual attributeswithin the dataset and achieves both the optimalnumber of clusters and the optimal classificationaccuracy. The validity of the proposed approach isinvestigated by comparing the classification resultsobtained for UCI datasets with those obtained bysupervised classification BPNN, decision-tree methods.There is good evidence to show that the proposedPFV-index method not only has a superior clusteringaccomplishment than the considered methods, but alsoachieves better classification accuracy.
Journal of Grey System(Volume 16)