Peter Hall received his BSc degree from the University of Sydney in 1974.  His MSc and DPhil
degrees are from the Australian National University and the University of Oxford, both in 1976.

He taught at the University of Melbourne before taking, in 1978, a position at the Australian National University, where
he worked until returning to Melbourne University in 2006.

His research interests range across a variety topics in probability and statistics.  He is a Fellow of the Australian Academy of Science and the Royal Society of London, and a winner of the COPSS Award.

Abstract

NONLINEAR METHODS FOR VARIABLE SELECTION

The conventional approach to variable selection, based on a linear model, can perform very effectively provided the response to relevant components is approximately monotone and its gradient changes only slowly. In other circumstances, nonlinearity of response can result in significant vector components being overlooked. Even if good results are obtained by linear model fitting, they can sometimes be bettered by using a nonlinear approach. These circumstances can arise in practice, with real data, and they motivate alternative methodologies. We suggest an approach based on ranking generalised empirical correlations between the response variable and components of the explanatory vector. This technique is not prediction-based, and can identify variables that are influential but not explicitly part of a linear predictive model.