Bounding the performance of statistical procedures using decision theory
開催期間
12:00 ~ 13:00
場所
講演者
概要
With the constant challenge of the need to analyse new data and to make full use of computational advances, statisticians are continually developing new methods of statistical inference. The performance of each new method needs to be rigorously assessed. Decision theory provides an important tool for this assessment. After providing an introduction to some of the basic concepts and techniques of decision theory, the new decision-theoretic results of Kabaila, Giri and Leeb (2010), Leeb and Kabaila (2016) and Kabaila and Kong (2016) will be described and their application to bounding the performance of confidence intervals will be discussed.
References:
Kabaila, P., Giri, K. and Leeb, H. (2010). Admissibility of the usual confidence interval in linear regression. Electronic Journal of Statistics.
Leeb, H. and Kabaila, P. (2016). Admissibility of the usual confidence set for the mean of a univariate or bivariate normal population: The unknown-variance case. Journal of the Royal Statistical Society, Series B.
Kabaila, P. and Kong, Y. (2016). Lower bounds on integrated risk, subject to inequality constraints. Australian & New Zealand Journal of Statistics.