Analysis of Poisson varying-coefficient models with autoregression

Taeyoung Park, Seonghyun Jeong

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

In the regression analysis of time series of event counts, it is of interest to account for serial dependence that is likely to be present among such data as well as a nonlinear interaction between the expected event counts and predictors as a function of some underlying variables. We thus develop a Poisson autoregressive varying-coefficient model, which introduces autocorrelation through a latent process and allows regression coefficients to nonparametrically vary as a function of the underlying variables. The nonparametric functions for varying regression coefficients are estimated with data-driven basis selection, thereby avoiding overfitting and adapting to curvature variation. An efficient posterior sampling scheme is devised to analyse the proposed model. The proposed methodology is illustrated using simulated data and daily homicide data in Cali, Colombia.

Original languageEnglish
Pages (from-to)34-49
Number of pages16
JournalStatistics
Volume52
Issue number1
DOIs
Publication statusPublished - 2018 Jan 2

Fingerprint

Varying Coefficient Model
Autoregression
Poisson Model
Regression Coefficient
Count
Serial Dependence
Latent Process
Varying Coefficients
Nonlinear Interaction
Overfitting
Autoregressive Model
Autocorrelation
Regression Analysis
Data-driven
Predictors
Siméon Denis Poisson
Time series
Likely
Curvature
Vary

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Park, Taeyoung ; Jeong, Seonghyun. / Analysis of Poisson varying-coefficient models with autoregression. In: Statistics. 2018 ; Vol. 52, No. 1. pp. 34-49.
@article{7e26369cc19f488ab00fbddc57a5971f,
title = "Analysis of Poisson varying-coefficient models with autoregression",
abstract = "In the regression analysis of time series of event counts, it is of interest to account for serial dependence that is likely to be present among such data as well as a nonlinear interaction between the expected event counts and predictors as a function of some underlying variables. We thus develop a Poisson autoregressive varying-coefficient model, which introduces autocorrelation through a latent process and allows regression coefficients to nonparametrically vary as a function of the underlying variables. The nonparametric functions for varying regression coefficients are estimated with data-driven basis selection, thereby avoiding overfitting and adapting to curvature variation. An efficient posterior sampling scheme is devised to analyse the proposed model. The proposed methodology is illustrated using simulated data and daily homicide data in Cali, Colombia.",
author = "Taeyoung Park and Seonghyun Jeong",
year = "2018",
month = "1",
day = "2",
doi = "10.1080/02331888.2017.1353514",
language = "English",
volume = "52",
pages = "34--49",
journal = "Statistics",
issn = "0233-1888",
publisher = "Taylor and Francis Ltd.",
number = "1",

}

Analysis of Poisson varying-coefficient models with autoregression. / Park, Taeyoung; Jeong, Seonghyun.

In: Statistics, Vol. 52, No. 1, 02.01.2018, p. 34-49.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Analysis of Poisson varying-coefficient models with autoregression

AU - Park, Taeyoung

AU - Jeong, Seonghyun

PY - 2018/1/2

Y1 - 2018/1/2

N2 - In the regression analysis of time series of event counts, it is of interest to account for serial dependence that is likely to be present among such data as well as a nonlinear interaction between the expected event counts and predictors as a function of some underlying variables. We thus develop a Poisson autoregressive varying-coefficient model, which introduces autocorrelation through a latent process and allows regression coefficients to nonparametrically vary as a function of the underlying variables. The nonparametric functions for varying regression coefficients are estimated with data-driven basis selection, thereby avoiding overfitting and adapting to curvature variation. An efficient posterior sampling scheme is devised to analyse the proposed model. The proposed methodology is illustrated using simulated data and daily homicide data in Cali, Colombia.

AB - In the regression analysis of time series of event counts, it is of interest to account for serial dependence that is likely to be present among such data as well as a nonlinear interaction between the expected event counts and predictors as a function of some underlying variables. We thus develop a Poisson autoregressive varying-coefficient model, which introduces autocorrelation through a latent process and allows regression coefficients to nonparametrically vary as a function of the underlying variables. The nonparametric functions for varying regression coefficients are estimated with data-driven basis selection, thereby avoiding overfitting and adapting to curvature variation. An efficient posterior sampling scheme is devised to analyse the proposed model. The proposed methodology is illustrated using simulated data and daily homicide data in Cali, Colombia.

UR - http://www.scopus.com/inward/record.url?scp=85025148206&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85025148206&partnerID=8YFLogxK

U2 - 10.1080/02331888.2017.1353514

DO - 10.1080/02331888.2017.1353514

M3 - Article

VL - 52

SP - 34

EP - 49

JO - Statistics

JF - Statistics

SN - 0233-1888

IS - 1

ER -