4.0 Article

Improved transformation-based quantile regression

Publisher

WILEY
DOI: 10.1002/cjs.11240

Keywords

Aranda-Ordaz transformation; Box-Cox transformation; Bounded response; two-stage estimation

Funding

  1. MRC [G0400546] Funding Source: UKRI
  2. Medical Research Council [G0400546] Funding Source: researchfish

Ask authors/readers for more resources

Modelling the quantiles of a random variable is facilitated by their equivariance to monotone transformations. In conditional modelling, transforming the response variable serves to approximate nonlinear relationships by means of flexible and parsimonious models; these usually include standard transformations as special cases. Transforming back to obtain predictions on the original scale or to calculate marginal nonlinear effects becomes a trivial task. This approach is particularly useful when the support of the response variable is bounded. We propose novel transformation models for singly or doubly bounded responses, which improve upon the performance of conditional quantile estimators as compared to other competing transformations, namely the Box-Cox and the Aranda-Ordaz transformations. The key is to provide flexible transformations with range the whole of the real line. Estimation is carried out by means of a two-stage estimator, while confidence intervals are obtained by bootstrap. A simulation study and some illustrative data analyses are presented. The Canadian Journal of Statistics 43: 118-132; 2015 (c) 2015 Statistical Society of Canada Resume L'equivariance aux transformations monotones des quantiles d'une variable aleatoire facilite leur modelisation. Dans le cadre d'une modelisation conditionnelle, une transformation de la variable reponse permet d'estimer des relations non lineaires a l'aide de modeles simples et parcimonieux. Les transformations envisagees incluent typiquement des transformations standards comme cas particuliers. La conversion de previsions vers l'echelle d'origine ou le calcul des effets marginaux non lineaires se font alors de facon triviale par l'application de la transformation inverse. Les auteurs proposent un modele de transformation novateur pour des reponses bornees unilateralement ou bilateralement. Les performances observees pour l'estimation conditionnelle de quantiles surpassent celles des transformations concurrentes comme la transformee de Box-Cox et celle d'Aranda-Ordaz. La cle consiste a proposer des transformations flexibles dont le domaine couvre tout l'axe reel. Les auteurs effectuent l'estimation a l'aide d'une methode en deux etapes et ils determinent les intervalles de confiance par le reechantillonnage. Ils presentent quelques simulations et des exemples d'analyse de donnees. La revue canadienne de statistique 43: 118-132; 2015 (c) 2015 Societe statistique du Canada

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.0
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

No Data Available
No Data Available