A NOVEL HYBRID CLUSTERING ALGORITHM : INTEGRATED PARTITIONAL AND HIERARCHICAL CLUSTERING ALGORITHM FOR CATEGORICAL DATA

Abstract

Data clustering became increasingly important in the field of computational statistics and data mining. Many algorithms have been developed in the literature for clustering where k-means clustering and hierarchical clustering are two well-known algorithms to partition the numerical data into groups. Due to the disadvantages of both categories of algorithms, recent researches have focused on hybrid clustering that combines the features of hierarchical and partitional clustering. The present paper developed a novel hybrid clustering algorithm called Integrated Partitional and Hierarchical clustering algorithm (IPHC) for categorical data. The proposed IPHC used modified k-mode clustering algorithm which is our previous proposed work for sub-clustering of categorical data and representative points (top-q points) are chosen from the sub-cluster. These representative points are then applied to agglomerative hierarchical clustering algorithm for constructing the hierarchical tree, called dendrogram. The proposed IPHC approach is validated with the aid of real categorical dataset available in the UCI machine learning repository. The experimental results demonstrate the effectiveness of the proposed IPHC algorithm.