作者: Peter Filzmoser , None
DOI:
关键词: Multivariate outlier detection 、 Mahalanobis distance 、 Data structure 、 Outlier 、 Measure (mathematics) 、 Sample size determination 、 Pattern recognition 、 Artificial intelligence 、 Computer science 、 Distribution function 、 Empirical distribution function
摘要: A method for the detection of multivariate outliers is proposed which accounts data structure and sample size. The cut-off value identifying defined by a measure deviation empirical distribution function robust Mahalanobis distance from theoretical function. easy to implement fast compute.