Seed clustering by using subspace clustering algorithm and feature selection algorithm

Cluster analysis of human-generated seed data exhibits inefficiency and lacks robust validation measures. This research focuses on identifying crucial cluster attributes through a combination of feature selection, Genetic Algorithm, and subspace clustering techniques. In today’s data-driven landscape, clustering plays a pivotal role in grouping data items into meaningful clusters, optimizing likeness within a cluster while reducing it among the clusters. Traditional algorithm for clustering employs the entirety of the feature space dimensions, often leading to suboptimal results due to irrelevant and redundant attributes. Author introduces two novel methods: GA-FS Clustering (utilizing Genetic Algorithm for feature selection) and COSA-Clustering (clustering based on a subset of attributes) tailored for clustering high-dimensional data. The performance of these methods is rigorously analyzed, aiming to enhance the efficiency and accuracy of cluster analysis in complex datasets.