The Other Measurement Error Problem

Aurore Delaigle

Department of Mathematics and Statistics, University of Melbourne

In many real life applications, observations can not be measured precisely and the only data available are contaminated by measurement errors (for example, due to the inaccuracy of the measurement device used).

When applied to such data, standard regression estimators are not consistent. A lot of attention in the literature has been devoted to the development techniques for regression estimation which are valid in the error case, and we will discuss a recent nonparametric solution to this problem. Another important problem which has been much less studied in the measurement error literature is the estimation of the "variance curve" in regression. We will show that the two problems are strongly related, and present some techniques for estimating the variance curve.

Aurore is a Principal Research Fellow/QEII Fellow in the Department of Mathematics and Statistics at the University of Melbourne. She completed her PhD at the Institute of Statistics, Université Catholique de Louvain in Belgium. She is currently on leave from her position as Associate Professor at the University of Bristol. She is an associate editor for the Journal of the American Statistical Association, the Journal of Computational and Graphical Statistics, Statistica Sinica and the Journal of the Korean Statistical Society. Aurore’s research interests include nonparametric curve estimation, measurement error, bandwidth selection and functional data analysis.

For information about the speaker, see http://www.ms.unimelb.edu.au/~aurored/