Discrete dispersion models and their Tweedie asymptotics

Bent Jørgensen, Célestin C. Kokonendji

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningpeer review

Resumé

The paper introduce a class of two-parameter discrete dispersion models, obtained by combining convolution with a factorial tilting operation, similar to exponential dispersion models which combine convolution and exponential tilting. The equidispersed Poisson model has a special place in this approach, whereas several overdispersed discrete distributions, such as the Neyman Type A, Pólya-Aeppli, negative binomial and Poisson-inverse Gaussian, turn out to be Poisson-Tweedie factorial dispersion models with power dispersion functions, analogous to ordinary Tweedie exponential dispersion models with power variance functions. Using the factorial cumulant generating function as tool, we introduce a dilation operation as a discrete analogue of scaling, generalizing binomial thinning. The Poisson-Tweedie factorial dispersion models are closed under dilation, which in turn leads to a Poisson-Tweedie asymptotic framework where Poisson-Tweedie models appear as dilation limits. This unifies many discrete convergence results and leads to Poisson and Hermite convergence results, similar to the law of large numbers and the central limit theorem, respectively. The dilation operator also leads to a duality transformation which in some cases transforms overdispersion into underdispersion and vice-versa. Finally, we consider the multivariate factorial cumulant generating function, and introduce a multivariate notion of over- and underdispersion, and a multivariate zero-inflation index.
Originalsprog Engelsk A St A - Advances in Statistical Analysis 100 1 43-78 1863-8171 https://doi.org/10.1007/s10182-015-0250-z Udgivet - 2016

Emneord

• discrete distribution
• factorial tilting family

Citer dette

Jørgensen, Bent ; Kokonendji, Célestin C. / Discrete dispersion models and their Tweedie asymptotics. I: A St A - Advances in Statistical Analysis. 2016 ; Bind 100, Nr. 1. s. 43-78.
@article{7323f5ba71d6457da4ceb96a5cf29aaf,
title = "Discrete dispersion models and their Tweedie asymptotics",
abstract = "The paper introduce a class of two-parameter discrete dispersion models, obtained by combining convolution with a factorial tilting operation, similar to exponential dispersion models which combine convolution and exponential tilting. The equidispersed Poisson model has a special place in this approach, whereas several overdispersed discrete distributions, such as the Neyman Type A, P{\'o}lya-Aeppli, negative binomial and Poisson-inverse Gaussian, turn out to be Poisson-Tweedie factorial dispersion models with power dispersion functions, analogous to ordinary Tweedie exponential dispersion models with power variance functions. Using the factorial cumulant generating function as tool, we introduce a dilation operation as a discrete analogue of scaling, generalizing binomial thinning. The Poisson-Tweedie factorial dispersion models are closed under dilation, which in turn leads to a Poisson-Tweedie asymptotic framework where Poisson-Tweedie models appear as dilation limits. This unifies many discrete convergence results and leads to Poisson and Hermite convergence results, similar to the law of large numbers and the central limit theorem, respectively. The dilation operator also leads to a duality transformation which in some cases transforms overdispersion into underdispersion and vice-versa. Finally, we consider the multivariate factorial cumulant generating function, and introduce a multivariate notion of over- and underdispersion, and a multivariate zero-inflation index.",
keywords = "discrete distribution, factorial tilting family",
author = "Bent J{\o}rgensen and Kokonendji, {C{\'e}lestin C.}",
year = "2016",
doi = "10.1007/s10182-015-0250-z",
language = "English",
volume = "100",
pages = "43--78",
journal = "A St A - Advances in Statistical Analysis",
issn = "1863-8171",
publisher = "Heinemann",
number = "1",

}

Discrete dispersion models and their Tweedie asymptotics. / Jørgensen, Bent; Kokonendji, Célestin C.

I: A St A - Advances in Statistical Analysis, Bind 100, Nr. 1, 2016, s. 43-78.

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningpeer review

TY - JOUR

T1 - Discrete dispersion models and their Tweedie asymptotics

AU - Jørgensen, Bent

AU - Kokonendji, Célestin C.

PY - 2016

Y1 - 2016

N2 - The paper introduce a class of two-parameter discrete dispersion models, obtained by combining convolution with a factorial tilting operation, similar to exponential dispersion models which combine convolution and exponential tilting. The equidispersed Poisson model has a special place in this approach, whereas several overdispersed discrete distributions, such as the Neyman Type A, Pólya-Aeppli, negative binomial and Poisson-inverse Gaussian, turn out to be Poisson-Tweedie factorial dispersion models with power dispersion functions, analogous to ordinary Tweedie exponential dispersion models with power variance functions. Using the factorial cumulant generating function as tool, we introduce a dilation operation as a discrete analogue of scaling, generalizing binomial thinning. The Poisson-Tweedie factorial dispersion models are closed under dilation, which in turn leads to a Poisson-Tweedie asymptotic framework where Poisson-Tweedie models appear as dilation limits. This unifies many discrete convergence results and leads to Poisson and Hermite convergence results, similar to the law of large numbers and the central limit theorem, respectively. The dilation operator also leads to a duality transformation which in some cases transforms overdispersion into underdispersion and vice-versa. Finally, we consider the multivariate factorial cumulant generating function, and introduce a multivariate notion of over- and underdispersion, and a multivariate zero-inflation index.

AB - The paper introduce a class of two-parameter discrete dispersion models, obtained by combining convolution with a factorial tilting operation, similar to exponential dispersion models which combine convolution and exponential tilting. The equidispersed Poisson model has a special place in this approach, whereas several overdispersed discrete distributions, such as the Neyman Type A, Pólya-Aeppli, negative binomial and Poisson-inverse Gaussian, turn out to be Poisson-Tweedie factorial dispersion models with power dispersion functions, analogous to ordinary Tweedie exponential dispersion models with power variance functions. Using the factorial cumulant generating function as tool, we introduce a dilation operation as a discrete analogue of scaling, generalizing binomial thinning. The Poisson-Tweedie factorial dispersion models are closed under dilation, which in turn leads to a Poisson-Tweedie asymptotic framework where Poisson-Tweedie models appear as dilation limits. This unifies many discrete convergence results and leads to Poisson and Hermite convergence results, similar to the law of large numbers and the central limit theorem, respectively. The dilation operator also leads to a duality transformation which in some cases transforms overdispersion into underdispersion and vice-versa. Finally, we consider the multivariate factorial cumulant generating function, and introduce a multivariate notion of over- and underdispersion, and a multivariate zero-inflation index.

KW - discrete distribution

KW - factorial tilting family

U2 - 10.1007/s10182-015-0250-z

DO - 10.1007/s10182-015-0250-z

M3 - Journal article

VL - 100

SP - 43

EP - 78

JO - A St A - Advances in Statistical Analysis

JF - A St A - Advances in Statistical Analysis

SN - 1863-8171

IS - 1

ER -