Scholarlink

<p>This paper introduces and quantifies the concept of overlap in matrix-variate data. The proposed methodology derives and computes the exact overlap, defined as the total misclassification probability between two clusters. Discriminant functions tailored for matrix-variate data are developed based on a specified level of overlap, and their effectiveness is evaluated. The performance of these classifiers is compared to that of traditional multivariate classifiers across various simulation scenarios, considering different overlap levels, sample sizes, and matrix dimensions. Results from both simulated and real-world datasets demonstrate the superior performance of discriminant functions based on the matrix distribution format.</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

Study on the Overlap in Matrix-Variate Data with Applications in Discriminant Analysis

  • 期刊论文

  • 发表日期：

    2025年7月28日

查看全文
  • Yingying Zhang,
  • Volodymyr Melnykov,
  • Xuwen Zhu

摘要

This paper introduces and quantifies the concept of overlap in matrix-variate data. The proposed methodology derives and computes the exact overlap, defined as the total misclassification probability between two clusters. Discriminant functions tailored for matrix-variate data are developed based on a specified level of overlap, and their effectiveness is evaluated. The performance of these classifiers is compared to that of traditional multivariate classifiers across various simulation scenarios, considering different overlap levels, sample sizes, and matrix dimensions. Results from both simulated and real-world datasets demonstrate the superior performance of discriminant functions based on the matrix distribution format.