A procedure for outlier identification in data from
continuous
distributions
ABSTRACT:
In this talk, I will introduce a procedure that can be used for
the identification of outliers in univariate or multivariate data
coming from continuous distributions. The limiting distribution of the
relevant statistic can be determined for completely specified models. In
many interesting situations, the finite-sample distribution, in the case
of estimated parameters, can be approximated by Monte Carlo simulation.
It will be shown that this procedure is very effective in the
identification of outliers through Monte Carlo simulations and further
that it is sensitive to sample size, a feature seldom found in outlier
identifiers.
About the Speaker
Prof. Balakrishnan did his B.Sc. and M.Sc. degrees at Madras
University in India. His PhD is from the Indian Institute of
Technology. He is currently a professor in the Department of Mathematics
and Statistics at McMaster University. His research interests include Order Statistics,
Inferential Methods, Distribution Theory, Life-testing and Reliability, Multivariate Analysis and Robust Inference. He is a Fellow of the American Statistical
Association and an Elected Member of the International Statistical Institute. Dr. Balakrishnan is on the editorial board of many journals including
Naval Research Logistics and Computational Statistics & Data Analysis and
is editor in chief of Communications in Statistics. Dr. Balakrishnan is
very well known for the large number of books and volumes that he has
written or edited as well as his invited talks around the world.
