Optimizing Textual Data Categorization: An Evaluation of K-Means, Mean Shift, and Hierarchical Clustering Methods

The study applies a dataset comprising of more than 20,000 documents to a comprehensive comparative analysis of clustering algorithms: Mean Shift, Hierarchical and K-Means. The study's main objective is to assess these algorithms' performance using the Davies-Bouldin Index, Calinski-Harabasz Index, and Silhouette Score. The efficiency of each algorithm in clustering high-dimensional text data is evaluated, exposing its distinct advantages and disadvantages. The most balanced technique is found to be K-Means clustering, which performs well in terms of cluster definition, separation, and density. This work contributes to a wider understanding of how various algorithms interpret and manage complex, high-dimensional data structures in addition to providing guidance for choosing suitable clustering techniques for similar datasets.