A SURVEY ON CLUSTERING METHODS FOR NUMERIC, CATEGORICAL AND MIXED VARIABLES DATA

Clustering is widely used in different fields such as biology, engineering, text mining, bioinformatics and agriculture. Most of the clustering methods use distance measures to find the similarity or dissimilarity between data objects. Traditional clustering algorithms use Euclidean distance measure to judge the similarity of two data objects. It works well when the attributes of a data set are purely quantitative. However, Euclidean distance measure fails to capture the similarity of data elements when attributes are qualitative or mixed. In this paper, clustering algorithms based on the data type containing numeric, categorical and mixed variables are reviewed such as DBSCAN, fuzzy c means, ROCK, K-modes, K-prototypes and Modified K-means.