Scholarlink

Class imbalance, where some clusters have significantly more curves than others, pose difficulties for functional clustering methods. Often, the largest clusters are prioritized at the expense of smaller clusters, resulting in a misclassification of curves. We propose functional iterative hierarchical clustering as a novel approach to cluster functional data with imbalanced classes. We compare our proposed method to existing functional clustering methods through a simulation study and assess the performance on a benchmark dataset. The proposed approach demonstrates improved clustering accuracy in terms of the adjusted rand index, achieving on average a 22% improvement over existing methods.

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

Functional Clustering with Imbalanced Groups

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Catherine Higgins,
  • Michelle Carey

摘要

Class imbalance, where some clusters have significantly more curves than others, pose difficulties for functional clustering methods. Often, the largest clusters are prioritized at the expense of smaller clusters, resulting in a misclassification of curves. We propose functional iterative hierarchical clustering as a novel approach to cluster functional data with imbalanced classes. We compare our proposed method to existing functional clustering methods through a simulation study and assess the performance on a benchmark dataset. The proposed approach demonstrates improved clustering accuracy in terms of the adjusted rand index, achieving on average a 22% improvement over existing methods.