A new bivariate negative binomial regression model

Pouya Faroughi, Noriszura Ismail

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper introduces a new form of bivariate negative binomial (BNB-1) regression which can be fitted to bivariate and correlated count data with covariates. The BNB regression discussed in this study can be fitted to bivariate and overdispersed count data with positive, zero or negative correlations. The joint p.m.f. of the BNB1 distribution is derived from the product of two negative binomial marginals with a multiplicative factor parameter. Several testing methods were used to check overdispersion and goodness-of-fit of the model. Application of BNB-1 regression is illustrated on Malaysian motor insurance dataset. The results indicated that BNB-1 regression has better fit than bivariate Poisson and BNB-2 models with regards to Akaike information criterion.

Original languageEnglish
Title of host publicationAIP Conference Proceedings
PublisherAmerican Institute of Physics Inc.
Pages732-736
Number of pages5
Volume1635
ISBN (Print)9780735412743
DOIs
Publication statusPublished - 2014
Event3rd International Conference on Quantitative Sciences and Its Applications: Fostering Innovation, Streamlining Development, ICOQSIA 2014 - Langkawi, Kedah
Duration: 12 Aug 201414 Aug 2014

Other

Other3rd International Conference on Quantitative Sciences and Its Applications: Fostering Innovation, Streamlining Development, ICOQSIA 2014
CityLangkawi, Kedah
Period12/8/1414/8/14

Fingerprint

regression analysis
goodness of fit
products

Keywords

  • Bivariate negative binomial regression
  • Bivariate Poisson regression
  • Correlation

ASJC Scopus subject areas

  • Physics and Astronomy(all)

Cite this

Faroughi, P., & Ismail, N. (2014). A new bivariate negative binomial regression model. In AIP Conference Proceedings (Vol. 1635, pp. 732-736). American Institute of Physics Inc.. https://doi.org/10.1063/1.4903663

A new bivariate negative binomial regression model. / Faroughi, Pouya; Ismail, Noriszura.

AIP Conference Proceedings. Vol. 1635 American Institute of Physics Inc., 2014. p. 732-736.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Faroughi, P & Ismail, N 2014, A new bivariate negative binomial regression model. in AIP Conference Proceedings. vol. 1635, American Institute of Physics Inc., pp. 732-736, 3rd International Conference on Quantitative Sciences and Its Applications: Fostering Innovation, Streamlining Development, ICOQSIA 2014, Langkawi, Kedah, 12/8/14. https://doi.org/10.1063/1.4903663
Faroughi P, Ismail N. A new bivariate negative binomial regression model. In AIP Conference Proceedings. Vol. 1635. American Institute of Physics Inc. 2014. p. 732-736 https://doi.org/10.1063/1.4903663
Faroughi, Pouya ; Ismail, Noriszura. / A new bivariate negative binomial regression model. AIP Conference Proceedings. Vol. 1635 American Institute of Physics Inc., 2014. pp. 732-736
@inproceedings{2e561dff09c24350817dc30414a76546,
title = "A new bivariate negative binomial regression model",
abstract = "This paper introduces a new form of bivariate negative binomial (BNB-1) regression which can be fitted to bivariate and correlated count data with covariates. The BNB regression discussed in this study can be fitted to bivariate and overdispersed count data with positive, zero or negative correlations. The joint p.m.f. of the BNB1 distribution is derived from the product of two negative binomial marginals with a multiplicative factor parameter. Several testing methods were used to check overdispersion and goodness-of-fit of the model. Application of BNB-1 regression is illustrated on Malaysian motor insurance dataset. The results indicated that BNB-1 regression has better fit than bivariate Poisson and BNB-2 models with regards to Akaike information criterion.",
keywords = "Bivariate negative binomial regression, Bivariate Poisson regression, Correlation",
author = "Pouya Faroughi and Noriszura Ismail",
year = "2014",
doi = "10.1063/1.4903663",
language = "English",
isbn = "9780735412743",
volume = "1635",
pages = "732--736",
booktitle = "AIP Conference Proceedings",
publisher = "American Institute of Physics Inc.",

}

TY - GEN

T1 - A new bivariate negative binomial regression model

AU - Faroughi, Pouya

AU - Ismail, Noriszura

PY - 2014

Y1 - 2014

N2 - This paper introduces a new form of bivariate negative binomial (BNB-1) regression which can be fitted to bivariate and correlated count data with covariates. The BNB regression discussed in this study can be fitted to bivariate and overdispersed count data with positive, zero or negative correlations. The joint p.m.f. of the BNB1 distribution is derived from the product of two negative binomial marginals with a multiplicative factor parameter. Several testing methods were used to check overdispersion and goodness-of-fit of the model. Application of BNB-1 regression is illustrated on Malaysian motor insurance dataset. The results indicated that BNB-1 regression has better fit than bivariate Poisson and BNB-2 models with regards to Akaike information criterion.

AB - This paper introduces a new form of bivariate negative binomial (BNB-1) regression which can be fitted to bivariate and correlated count data with covariates. The BNB regression discussed in this study can be fitted to bivariate and overdispersed count data with positive, zero or negative correlations. The joint p.m.f. of the BNB1 distribution is derived from the product of two negative binomial marginals with a multiplicative factor parameter. Several testing methods were used to check overdispersion and goodness-of-fit of the model. Application of BNB-1 regression is illustrated on Malaysian motor insurance dataset. The results indicated that BNB-1 regression has better fit than bivariate Poisson and BNB-2 models with regards to Akaike information criterion.

KW - Bivariate negative binomial regression

KW - Bivariate Poisson regression

KW - Correlation

UR - http://www.scopus.com/inward/record.url?scp=84916599271&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84916599271&partnerID=8YFLogxK

U2 - 10.1063/1.4903663

DO - 10.1063/1.4903663

M3 - Conference contribution

AN - SCOPUS:84916599271

SN - 9780735412743

VL - 1635

SP - 732

EP - 736

BT - AIP Conference Proceedings

PB - American Institute of Physics Inc.

ER -