Please use this identifier to cite or link to this item:
Author(s): Rui Jorge Gonçalves
Title: The Power Normal distribution
Issue Date: 2019-10-18
Abstract: The Inverse Box-Cox (BC) transformation (see [1]) produces the Power Normal (PN) distribution family that includes the log-normal and the normal distributions, see [2], [4] and [3]. The Box-Cox power transformation aims to transform data to approximate normality (truncated normal (TN)) therefore the knowledge of the its inverse scale is of major importance. In this paper we consider the univariate PN distribution, and, because in many applications there are more than one variable and correlation we also consider the bivariate PN distribution (BPN), see [6] and [5]. We give some important results concerning both distributions. In [6] it is given a formula to approximate the ordinary moments of the PN distribution, herein, we give an exact formula for the ordinary integer moments of the PN distribution. Regarding the BPN distribution, we calculate the marginal probability density functions (pdf) and we note that they are not univariate PN distributed however the conditional pdf is PN univariate. This is also true for the corresponding functions of the transformed scale, the TN distribution. The correlation curve or curve of the conditional moments E(X 2|X 1 = x 1) that is useful to characterize the structure of the correlation is very well fitted by a power law model f(x) = axb + c. (c) Published under licence by IOP Publishing Ltd.
Subject: Matemática, Matemática
Mathematics, Mathematics
Scientific areas: Ciências exactas e naturais::Matemática
Natural sciences::Mathematics
Source: International Conference on Mathematical Models & Computational Techniques in Science & Engineering , London, UK
Document Type: Artigo em Livro de Atas de Conferência Internacional
Rights: restrictedAccess
Appears in Collections:FEUP - Artigo em Livro de Atas de Conferência Internacional

Files in This Item:
File Description SizeFormat 
  Restricted Access
Texto integral84.59 kBAdobe PDF    Request a copy from the Author(s)

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.