An Agglomerative Fuzzy K-means Approach to Building Decision Cluster Classifiers

Classification is an important task in machine learning and data mining. Lots of classification models have been proposed based on different theories and assumptions. Several researchers have proposed to build classifiers by using a sequence of nested clusterings based on the assumption that if objects are spatially close to one another in data space, they tend to have the same categorical label. In this paper, we pro- pose to build such classifiers by using the agglomerative fuzzy k-means because of its three good properties: (1) the clustering algorithm is insensitive to initial centers; (2) the algorithm can automatically determine the number of clusters combined with cluster validation techniques; (3) the algorithm can control the density level of clusters identified. The comparison experiments with traditional classifiers on benchmark data sets from UCI Machine Learning Repository have shown the effectiveness of our proposed approach.