Multi-Objective Scatter Search using Cuckoo Search with Modified Hierarchical Agglomerative Clustering for Resolving Clustering Problems

Data clustering is a vital concept of mining as it partitions the given dataset into meaningful set of clusters based on data similarity. This concept enhances the computation efficiency in the data analysis processes. In the preceding researches, Particle Swarm Artificial Bee Colony (PSABC) and Hybrid Artificial Bee Colony-Firefly Algorithm (HABC-FA) were developed for clustering to solve the clustering problem; however the imbalanced dataset problem still prevailed. This was overcome by the introduction of Multi-Objective Scatter Search Simulated Annealing with Hierarchical Agglomerative Clustering (MOSSSA-HAC) approach. But the feature selection in this approach has slow convergence speed. Hence in this paper, the Multi-Objective Scatter Search with Cuckoo Search algorithm with Modified Hierarchical Agglomerative Clustering (MOSSCS-MHAC) is proposed for resolving the convergence problem. In the initial stage, the Modified KNN is used for pre-processing followed by the under sampling process for error reduction. Then the multi-objective feature selection is performed using MOSSCS while the final clustering is achieved using MHAC algorithm. This approach provides optimal clustering with high accuracy. The experimental results show that the proposed MOSSCS-MHAC provides high values of precision, recall and f-measure than the existing algorithms.