Enhanced Unsupervised Feature Selection Method Using Crow Search Algorithm and Calinski-Harabasz

This paper proposes enhancing the K-means clustering method by incorporating the Crow Search Algorithm (CSA) and Calinski-Harabasz (CH) index to address the issue of determining the optimal number of clusters and attribute selection. The proposed approach, called Crow Search Algorithm K-mean clustering (CSAK_means), aims to explore the search space more effectively to find the best solutions. The efficiency of the CSAK_means algorithm is evaluated using a comparative experimental study for five datasets from the UCI repositories: Wine, Bodega, Cmc, Zoo, and Abalone. The results confirm that the proposed method outperforms the default algorithms in terms of average feature selection performance and silhouette value.