Robust hotelling's statistic for test a hypothesis of mean multivariate population based on RMCD

Abstract EN

Hotelling’s statistic 2 T is an important to test a hypothesis about the mean of a multivariate normal population with location and scale parameters (with the proposition that the scale parameter is unknown positive definite matrix ) .

However, hypothesis test based on this statistic can be adversely affected by outliers.

In this paper, an alternative technique is proposed based on a statistic which uses the reweighted minimum covariance determinant (RMCD) estimators instead of the classical mean vector and covariance matrix .

A simulation technique has been used as a technique to make a comparison between the classical and the proposed statistic by generating the data that have a contaminated multivariate normal distribution from one side and from two sides .

The results have shown that, the proposed robust statistic is almost bett.