The metrically trimmed mean as a robust estimator of location

摘要: The metrically trimmed mean is defined as the average of observations remaining after a fixed number outlying have been removed. A metric, distance from median, used to determine which points are outlying. influence curve and asymptotic normality derived using von Mises expansions. relative merits discussed in neighborhoods nonparametric models with natural parameters. It observed that works well for center symmetry symmetric distribution function asymmetric contamination. multivariate extension discussed.