Scholarlink

The detection of outliers is essential for ensuring the integrity and accuracy of data analysis. This study highlights the importance of measuring the performance of outlier detection methods and tests using robust criteria and performance measures. By applying the criteria of David and Paulson, we evaluate several discordance tests as well as popular outlier detection methods. We introduce five performance measures ( \(P_{1}\) , \(P_{2}\) , \(P_{3}\) , \(P_{4}\) , \(P_{5}\) ) to compare the efficiency of each method. Additionally, our novel contribution includes the use of the measure \(P_{2}\) to specifically evaluate outlier filtering methods. The results of this study provide valuable insights for selecting the most reliable and effective techniques in various analytical contexts, thereby ensuring robust and precise data analysis.

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

Comparison of Discordance Tests and Outlier Filtering Methods Based on David and Paulson’s Performance Measures

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kamilia Haddadou,
  • Lynda Atil,
  • Hocine Fellag

摘要

The detection of outliers is essential for ensuring the integrity and accuracy of data analysis. This study highlights the importance of measuring the performance of outlier detection methods and tests using robust criteria and performance measures. By applying the criteria of David and Paulson, we evaluate several discordance tests as well as popular outlier detection methods. We introduce five performance measures ( \(P_{1}\) , \(P_{2}\) , \(P_{3}\) , \(P_{4}\) , \(P_{5}\) ) to compare the efficiency of each method. Additionally, our novel contribution includes the use of the measure \(P_{2}\) to specifically evaluate outlier filtering methods. The results of this study provide valuable insights for selecting the most reliable and effective techniques in various analytical contexts, thereby ensuring robust and precise data analysis.