A Bayesian Analysis of Zero-Inflated Count Data: An Application to Youth Fitness Survey

Author

Liying Lu ; Yingzi Fu ; Peixiao Chu ; Xiaolin Zhang

Author_Institution

Fac. of Sci., Kunming Univ. of Sci. & Technol., Kunming, China

fYear

2014

fDate

15-16 Nov. 2014

Firstpage

699

Lastpage

703

Abstract

Count data with excess zeroes are widely encountered in the fields of biomedical, medical, insurance and public health survey. Zero-inflated Poisson (ZIP) regression model is an useful tool to analyze such data. In this paper, a Bayesian analysis approach is proposed, where the data augmentation strategy in combination with Gibbs sampler and M-H algorithm is used to obtain the Bayesian estimates of model parameters, moreover, DIC criterion as well as Posterior Prediction P-value (ppp-value) are also considered for model selection and for assessing the goodness-of-fit of the proposed model. Finally, a real example from youth fitness survey is adapted to illustrate the application of our methodology.

Keywords

Bayes methods; data handling; regression analysis; stochastic processes; Bayesian analysis; DIC criterion; Gibbs sampler; M-H algorithm; Posterior Prediction P-value; ZIP regression model; data augmentation strategy; excess zeroes; model selection; ppp-value; public health survey; youth fitness survey; zero-inflated Poisson regression model; zero-inflated count data; Analytical models; Bayes methods; Computational modeling; Data models; Estimation; Predictive models; Standards; Bayesian model selection; Posterior Prediction P-value; count data; zero-inflation;

fLanguage

English

Publisher

ieee

Conference_Titel

Computational Intelligence and Security (CIS), 2014 Tenth International Conference on

Conference_Location

Kunming

Print_ISBN

978-1-4799-7433-7

Type

conf

DOI

10.1109/CIS.2014.125

Filename

7016987