Three Case Studies Using Agglomerative Clustering

• Published in: Intelligent Systems Design and Applications Vol. 557; pp. 67 - 76
• Main Authors: Camargos, Rodrigo C., do Carmo Nicoletti, Maria
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2017; Springer International Publishing
• Series: Advances in Intelligent Systems and Computing
• Subjects: Agglomerative clustering; Artificial intelligence; Case studies in clustering; Unsupervised machine learning
• ISBN: 9783319534794; 3319534793
• ISSN: 2194-5357; 2194-5365
• DOI: 10.1007/978-3-319-53480-0_7

Finding a data clustering in a data set is a challenging task since algorithms usually depend on the adopted inter-cluster distance as well as the employed definition of cluster diameter. The work described in this paper approaches a well-known agglomerative clustering algorithm named AGNES (Agglomerative Nesting), in regards to its performance on three case studies namely, datasets formed by clusters of different sizes, uneven inter-cluster distances and diameters. Clustering results are evaluated using three well-known indexes, Dunn, Davies-Bouldin and Rand. Results obtained with K-means were used for comparison purposes. The experiments were conducted divided into three case studies. Their results suggest that AGNES and K-means have similar performance as far as identifying clusters with different sizes and inter-cluster distances, however, AGNES obtained the best results when dealing with clusters having both, different sizes and diameters.