Thesis, Undergraduate (BS Applied Mathematics)-U.P. Mindanao

Clustering is the assignment of a set of observations into subsets, known as clusters, so that observations in the same cluster are similar in some sense and observations in different clusters are dissimilar in the same sense. Firefly algorithm is a metaheuristic algorithm, inspired by the flashing behavior of fireflies, which operates through the use of a firefly's flash acting as a signal system to attract other fireflies. Although the firefly algorithm was found to be promising in optimization problems, its performance in clustering problems is still not known. Thus, this study, a clustering technique base on the firefly algorithm was formulated. Its effectiveness in clustering data sets was based on the quantization error. In this study, different values were tested for the random step size alpha for its parameter settings; it was found out that the quantization error decreases if the value of alpha is increased. Results showed that although the firefly algorithm performed better than the particle swarm optimization, it was not able to generate a quantization error lower than that of multi-elitist particle swarm optimization-tabu search. In order to improve firefly algorithm, further studies are still needed to explore the potential of firefly algorithm in data clustering.