초기 모드 결정 방식을 개선한 K-mode 알고리즘

Abstract

Data mining is the process of uncovering previously unknown patterns and relationships in large databases using sophisticated statistical analysis and modeling techniques such as classification, association rule mining, clustering, etc.. Spacially, clustering is an important data mining problem. Clustering, in data mining, is useful for discovering groups and identifying interesting distributions in underlying data. The k-means algorithm is well known for its efficiency in clustering large data sets. However, working only on numeric values prohibits it from being used to cluster real world data containing categorical values. Huang presented an algorithm, called K-mode algorithm, to extend the K-means paradigm to categorical domains(1997). K-mode algorithm suffers from initial staring conditions effect (initial mode, the number of initial mode). This paper improved the problem of K-mode algorithm using Max-Min method that is a kind of methods to decide initial values in K-means algorithm. We introduce new similarity measures to deal with categorical data sets using means of cluster. Tested with the Mushroom data sets and Small Soybean data sets the proposed algorithm has shown a good performance for the two aspects (accuracy, run time).