Combinations of the Greedy Heuristic Method for Clustering Problems and Local Search Algorithms

Abstract

In this paper, we investigate application of various options of algorithms with greedy agglomerative heuristic procedure for object clustering problems in continuous space in combination with various local search methods. We propose new modifications of the greedy agglomerative heuristic algorithms with local search in SWAP neighborhood for the p-medoid problems and j-means procedure for continuous clustering problems (p-median and k-means). New modifications of algorithms were applied to clustering problems in both continuous and discrete settings. Computational results with classical data sets and real data show the comparative efficiency of new algorithms for middle-size problems only.