Scholarlink

<p>Clustering algorithms have long sought to replicate human expertise in data clustering. This research utilizes six mechatronic products and a team of domain experts to assess how closely the results of six clustering algorithms align with expert-generated outcomes. This comparison of clustering results includes validation indices, the analysis of component migrations between the generated clusters, and optical inspection. The mechatronic products were selected for their complex nature, which provides a challenging context for the clustering analysis. The algorithms used in this research are Ward’s method (WARD), partitioning around medoids (PAM), K-means (K-MEANS), divisive analysis (DIANA), density-based spatial clustering of applications with noise (DBSCAN), and clustering by fast search and find of density peaks (DPC), and the validation indices employed to evaluate the clustering results are the silhouette coefficient (SC) and the composed density between and within clusters (CDbw).</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Systematic Evaluation of Clustering Algorithms Against Expert-Derived Clustering

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年4月24日

查看全文
  • Ioannis Mikrou,
  • Nickolas S. Sapidis

摘要

Clustering algorithms have long sought to replicate human expertise in data clustering. This research utilizes six mechatronic products and a team of domain experts to assess how closely the results of six clustering algorithms align with expert-generated outcomes. This comparison of clustering results includes validation indices, the analysis of component migrations between the generated clusters, and optical inspection. The mechatronic products were selected for their complex nature, which provides a challenging context for the clustering analysis. The algorithms used in this research are Ward’s method (WARD), partitioning around medoids (PAM), K-means (K-MEANS), divisive analysis (DIANA), density-based spatial clustering of applications with noise (DBSCAN), and clustering by fast search and find of density peaks (DPC), and the validation indices employed to evaluate the clustering results are the silhouette coefficient (SC) and the composed density between and within clusters (CDbw).