School of Mathematical Sciences

External Seminar: Han Lin Shang (Australian National University)

Location
Maths A17
Date(s)
Thursday 17th May 2018 (14:00-15:00)
Contact
Gilles Stupfler
Description

[Statistics & Probability Seminar]

Hypothesis tests for variable and model selections in scalar-on-function regression

We consider a likelihood ratio test to determine the optimal number of derivatives for a functional covariate in a linear scalar-on-function regression. When some real-valued covariates are included, we consider a regression equation specification error test (RESET) to examine the statistical significance of the real-valued covariates. Incorporating a possible nonlinear relationship, we apply a J test to determine the statistical significance of the nonlinear structure between a functional covariate and a scalar response. Using a simulation study and a chemometric data set, we study their finite-sample performance. (Joint work with Giles Hooker at Cornell University)

 

School of Mathematical Sciences

The University of Nottingham
University Park
Nottingham, NG7 2RD

For all enquiries please visit:
www.nottingham.ac.uk/enquire