Fractional Lion Algorithm-An Optimization Algorithm for Data Clustering

Abstract

Clustering divides the data available as bulk into meaningful, useful groups (Clusters) without any prior knowledge about the data. Cluster analysis provides an abstraction from individual data objects to the clusters in which those objects reside. It is a key technique in the data mining and has become an important issue in many fields. This paper presents a novel Fractional Lion Algorithm (FLA) as an optimization methodology for the clustering problems. The proposed algorithm utilizes the lion's unique characteristics such as pride, laggardness exploitation, territorial defence and territorial take over. The Lion algorithm is modified with the fractional theory to search the cluster centroids. The proposed fractional lion algorithm estimates the centroids with the systematic initialization itself. Proposed methodology is a robust one, since the parameters utilized are insensitive and not problem dependent. The performance of the proposed rapid centroid estimation is evaluated using the cluster accuracy, jaccard coefficient and rand coefficient. The quality of this approach is evaluated on the benchmarked iris and wine data sets. On comparing with the particle swarm clustering algorithm, experimental results shows that the clustering accuracy of about 75% is achieved by the proposed algorithm.