Scholarlink

Highest-Density Regions (HDRs) allow uncertainty estimation of predictions, estimates, or distributions of interest by identifying the samples with the highest density while covering the smallest possible volume. Therefore, they ensure the greatest efficiency. This paper extends the framework proposed by [3] to estimate HDRs using neighborhood measures. We generalize their approach to a multivariate setting (with the dimension \(d \ge 3\) ), using both nonparametric distance-based measures and parametric measures making use of vine copulas.

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

Using Vine Copulas for Estimating Highest-Density Regions in Multivariate Data

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Emanuele Masillo,
  • Nina Deliu

摘要

Highest-Density Regions (HDRs) allow uncertainty estimation of predictions, estimates, or distributions of interest by identifying the samples with the highest density while covering the smallest possible volume. Therefore, they ensure the greatest efficiency. This paper extends the framework proposed by [3] to estimate HDRs using neighborhood measures. We generalize their approach to a multivariate setting (with the dimension \(d \ge 3\) ), using both nonparametric distance-based measures and parametric measures making use of vine copulas.