The AVAS (Additivity And Variance Stabilization) algorithm of Tibshirani provides a non-parametric transformation of the response in a linear model to approximately constant variance. It is thus a generalization of the much used Box-Cox transformation. However, AVAS is not robust. Outliers can have a major effect on the estimated transformations both of the response and of the transformed explanatory variables in the Generalized Additive Model (GAM).We describe and illustrate robust methods for the non-parametric transformation of the response and for estimation of the terms in the model and report the results of a simulation study comparing our robust procedure with AVAS. We illustrate the efficacy of our procedure through a simulation study and the analysis of real data.

Robust response transformations for generalized additive models via additivity and variance stabilisation / Riani, Marco; Atkinson, Anthony; Corbellini, Aldo. - (2022).

Robust response transformations for generalized additive models via additivity and variance stabilisation

Riani Marco;Atkinson Anthony;Corbellini Aldo
2022-01-01

Abstract

The AVAS (Additivity And Variance Stabilization) algorithm of Tibshirani provides a non-parametric transformation of the response in a linear model to approximately constant variance. It is thus a generalization of the much used Box-Cox transformation. However, AVAS is not robust. Outliers can have a major effect on the estimated transformations both of the response and of the transformed explanatory variables in the Generalized Additive Model (GAM).We describe and illustrate robust methods for the non-parametric transformation of the response and for estimation of the terms in the model and report the results of a simulation study comparing our robust procedure with AVAS. We illustrate the efficacy of our procedure through a simulation study and the analysis of real data.
Robust response transformations for generalized additive models via additivity and variance stabilisation / Riani, Marco; Atkinson, Anthony; Corbellini, Aldo. - (2022).
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11381/2933992
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact