An Effective Data Clustering System using Weighted K-Means and Firefly Optimization Algorithms

Abstract

Clustering of data is a standard way used for analyzing the data in several applications such as, data mining, image analysis, pattern recognition, etc. The weighted K-means clustering is one amongst the various data mining techniques used for clustering of the data. The key advantages of weighted k-means clustering are efficient in managing huge amount of data, easy to implement, scalable, simple and easily modifiable. In contrast, the major disadvantage of weighted K-means clustering is the problem with choosing the initial centroids. This clustering technique chooses the initial centroids randomly that leads to a local optimum solution. To address this concern, an effective naturally-inspired optimization algorithm: fire-fly optimization is combined with weighted k-means clustering for obtaining the global optimum solution. In this research paper, weighted k-means clustering along with fire-fly optimization algorithm was developed for enhancing the performance of information sharing and searching efficiency among the population. Here, the proposed system was experimented on dissimilar medical datasets such as, heart disease (original), heart disease (stat-log), liver disease and Indian liver patients. In the practical study, the proposed method enhances the performance up to 0.02-0.4 (label value) as compared to the existing systems by using the concept of precision, recall, and FB-cubed.