Supplementary material to the paper
Marco Riani | Anthony C. Atkinson | Aldo Corbellini |
Department of Economics and Managemenet and Interdepartmental Centre for Robust Statistics | The London School of Economics, | Department of Economics and Managemenet and Interdepartmental Centre for Robust Statistics |
University of Parma | London WC2A 2AE, UK | University of Parma |
Italy | UK | Italy |
mriani@unipr.it | a.c.atkinson@lse.ac.uk | aldo.corbellini@unipr.it |
Abstract
The paper introduces an automatic procedure for the
transformation of the response in regression models to approximate normality.
Because incorrectly transformed response can generate spurious outliers,
robustness is essential. We consider the Box-Cox transformation and its
generalization to the extended Yeo-Johnson transformation which allows for
responses of both signs. The usefulness of our automatic procedure is
demonstrated on four sets of data. An important theoretical development is an
extension of the Bayesian Information Criterion (BIC) to allow for the
comparison of models following the deletion of observations, the number deleted
depending on the transformation parameter.
Data used in the paper are included in release 2020A of the FSDA toolbox downloadable from
Mathworks file exchange
https://www.mathworks.com/matlabcentral/fileexchange/72999-fsda
or from github repo
https://github.com/UniprJRC/FSDA
Matlab code used in the paper
The matlab file which creates all the figures can be downloaded here
Last modified 27/08/2020 11.05.45