Cookies

We use cookies to ensure that we give you the best experience on our website. You can change your cookie settings at any time. Otherwise, we'll assume you're OK to continue.

Department of Mathematical Sciences

Seminar Archives

On this page you can find information about seminars in this and previous academic years, where available on the database.

Statistics Seminars: Statistics with imprecise data

Presented by Scott Ferson, University of Liverpool

13 February 2017 14:00 in CM221

Statistics, as a discipline, has spent the last 100 years developing
methods for analyses in which sample size of data sets is limiting.
But samples sizes are not small anymore, as exemplified in financial
data, continuous mechanized measurements, satellite imagery and other
mass collections, commercial data, social media, and the coming
Internet of Things. Although sample size will always be an issue, it
may not the only issue or even the main issue confronting
statisticians, as other concerns become relatively more important as
sample sizes grow. These concerns include measurement imprecision,
missingness, censoring, biases, model uncertainties, nonstationarity,
and even nonrandomness. These concerns can be addressed by the
theory of imprecise probabilities which begs the existence of an
analogous *statistics of imprecise data*. We explore several
alternative approaches to handling intervals as native data
structures and show their advantages over currently popular
approaches to data censoring that make untenable assumptions about
the measurement process and therefore lead to grossly misleading
results that may not approach the correct answers even with
infinitely many samples.

Contact sunil.chhita@durham.ac.uk for more information