Outlier detection for high-dimensional data

• 成果类型: Article
• 署名作者: Ro, Kwangil; Zou, Changliang; Wang, Zhaojun; Yin, Guosheng
• 署名单位: Nankai University; University of Hong Kong
• 刊物名称: BIOMETRIKA
• ISSN/ISSBN: 0006-3444
• DOI: 10.1093/biomet/asv021
• 发表日期: 2015
• 页码: 589599
• 关键词: multivariate; algorithm; estimator

摘要

Outlier detection is an integral component of statistical modelling and estimation. For high-dimensional data, classical methods based on the Mahalanobis distance are usually not applicable. We propose an outlier detection procedure that replaces the classical minimum covariance determinant estimator with a high-breakdown minimum diagonal product estimator. The cut-off value is obtained from the asymptotic distribution of the distance, which enables us to control the Type I error and deliver robust outlier detection. Simulation studies show that the proposed method behaves well for high-dimensional data.