GRADE: A Novel Gravitational Density-Based Clustering Approach in the Multi-Objective Framework

Background: Clustering analysis plays a vital role in obtaining knowledgeable data from the huge amount of data sets in knowledge discovery. Most of the traditional clustering algorithms do not work well with high dimensional data. The objective of effective clustering is to obtain well connected, compact, and separated clusters. Density-Based Clustering (DBSCAN) is one of the popular clustering algorithms that use local density information of data points to detect clusters with arbitrary shapes. The Gravitational Search Algorithm (GSA) is one of the effective approaches inspired by Newton’s law of gravitational force where every particle in the universe attracts every other particle with a force. Objective: The primary aim of this paper is to design and develop a novel multi-objective clustering approach to produce the desired number of valid clusters. Further, these resulting clusters are to be optimized to obtain an optimal solution. Methods: In the proposed approach, a hybrid clustering algorithm based on GSA along with DBSCAN is recommended to group the data into the desired number of clusters, and in the next phase of the algorithm, Particle swarm optimization technique is applied in order to optimize the solutions using the fitness functions. Results: In the analysis of the result, we employed two objective functions namely quantization error and inter–cluster distance on four real-life data sets such as Iris, Wine, Wisconsin, and Yeast to evaluate the performance of our algorithm. Conclusion: The effectiveness of the GRADE algorithm is comprehensively demonstrated by comparing it with the well-known traditional K-mean algorithm in terms of accuracy and computational time.