A Novel Approach for Finding Alternative Clusterings Using Feature Selection

Abstract

Alternative clustering algorithms target finding alternative groupings of a dataset, on which traditional clustering algorithms can find only one even though many alternatives could exist. In this research, we propose a method for finding alternative clusterings of a dataset based on feature selection. Using the finding that each clustering has a set of so-called important features, we find the possible important features for the altenative clustering in subsets of data; we transform the data by weighting these features so that the original clustering will not likely to be found in the new data space. We then use the incremental K-means algorithm to directly maximizes the quality of the new clustering found in the new data space. We compare our approach with some previous works on a collection of machine learning datasets and another collection of documents. Our approach was the most stable one as it resulted in different and high quality clusterings in all of the tests. The results showed that by using feature selection, we can improve the dissimilarity between clusterings, and by directly maximizing the clustering quality, we can also achieve better clustering quality than the other approaches.