Table of Contents#### Download Safari Books Online apps: Apple iOS | Android | BlackBerry

Entire Site

Free Trial

Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.

240 Introduction to Robust Estimation and Hypothesis Testing are supplied, which eliminate the need to use the R command library(MASS) in order to access the R functions cov.mve and cov.mcd. Both of these functions return weights indicating whether a point is declared an outlier using the MVE and MCD methods. (It is noted that the S-PLUS version of the MVE and MCD outlier detection methods can give results that differ from the R versions used here.) However, these functions do not return the MVE and MCD estimate of location and scatter, but rather a W-estimate of location and scatter. (See Section 6.5.) In essence, points declared outliers are removed and the mean and covariance matrix are computed using the data that remain. R reports which subset of half of the data was used to compute the MVE and MCD estimates of location. So it is possible to determine the MVE and MCD estimates of location if desired. In some situations it is convenient to have an R function that returns just the MVE measure of location. Accordingly, the R function mvecen(m) is supplied to accomplish this goal. The R function mcdcen(m) computes the MCD measure of location.