Clustering is a popular data analysis and mining technique. A popular technique for clustering is based on k-means such that the data is partitioned into K clusters. However, the k-means algorithm highly depends on the initial state and converges to local optimum. The existing work presents a hybrid evolutionary algorithm to solve nonlinear partitional clustering problem. The evolutionary algorithm is the combination of FAPSO (fuzzy adaptive particle swarm optimization), ACO (ant colony

more »

... tion) and k-means algorithms, called FAPSO-ACO-K, which can find better cluster partition. Then k-means clustering is applied to get cluster results. K-means clustering is sensitive to the outliers and a set of objects closest to a centroid may be empty, in which case centroids cannot be updated. In k-means difficult to predict K-Value and different initial partitions can result in different final clusters. The objective of the proposed work is to overcome these problems, the K-medoids clustering algorithm where representative objects called medoids are considered instead of centroids. Because it uses the most centre located object in a cluster. The algorithm has excellent feature which requires the distance between every pairs of objects only once and uses this distance at every iterative step. It is less sensitive to outliers compared with the K-means clustering. It gives better performance than K-means clustering. Minimize the sensitivity of k-means to outliers. Pick the actual objects to represent clusters instead of mean values. Each remain object is clustered with the representative object (Medoid) to which is the most similar. The performance of the proposed work is evaluated through several benchmark data sets. The simulation result shows that the performance of the proposed work is better than the existing algorithm in terms of accuracy, recall, precision and F-measure.