The spatio-temporal multivariate Poisson lognormal model

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

Abstract

To deal with the variation and correlation structure of accident data along with recognized covariate effects, we develop a spatio-temporal model for multivariate accident count data. Based on the multivariate Poisson lognormal model, we introduce linear combinations of random impulses to capture spatial correlation. For temporal effects, the lagged observations are added to this model. Model estimation is carried out using Markov Chain Monte Carlo methods. Simulated data sets are used in assessing the performance of this model. An advantage of this new model is that it not only copes with three sources of variations; time, space and multivariate data variations, but also provides information on time and space dependency. The model is also capable of providing an improvement in the accuracy of count data modelling.

Original languageEnglish
Title of host publicationProceeding of the 25th National Symposium on Mathematical Sciences, SKSM 2017
Subtitle of host publicationMathematical Sciences as the Core of Intellectual Excellence
PublisherAmerican Institute of Physics Inc.
Volume1974
ISBN (Electronic)9780735416819
DOIs
Publication statusPublished - 28 Jun 2018
Event25th National Symposium on Mathematical Sciences: Mathematical Sciences as the Core of Intellectual Excellence, SKSM 2017 - Kuantan, Pahang, Malaysia
Duration: 27 Aug 201729 Aug 2017

Other

Other25th National Symposium on Mathematical Sciences: Mathematical Sciences as the Core of Intellectual Excellence, SKSM 2017
CountryMalaysia
CityKuantan, Pahang
Period27/8/1729/8/17

Fingerprint

accidents
Markov chains
Monte Carlo method
impulses

Keywords

  • count model
  • multivariate
  • traffic accident

ASJC Scopus subject areas

  • Physics and Astronomy(all)

Cite this

Zamzuri, Z. H. (2018). The spatio-temporal multivariate Poisson lognormal model. In Proceeding of the 25th National Symposium on Mathematical Sciences, SKSM 2017: Mathematical Sciences as the Core of Intellectual Excellence (Vol. 1974). [020013] American Institute of Physics Inc.. https://doi.org/10.1063/1.5041544

The spatio-temporal multivariate Poisson lognormal model. / Zamzuri, Zamira Hasanah.

Proceeding of the 25th National Symposium on Mathematical Sciences, SKSM 2017: Mathematical Sciences as the Core of Intellectual Excellence. Vol. 1974 American Institute of Physics Inc., 2018. 020013.

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

Zamzuri, ZH 2018, The spatio-temporal multivariate Poisson lognormal model. in Proceeding of the 25th National Symposium on Mathematical Sciences, SKSM 2017: Mathematical Sciences as the Core of Intellectual Excellence. vol. 1974, 020013, American Institute of Physics Inc., 25th National Symposium on Mathematical Sciences: Mathematical Sciences as the Core of Intellectual Excellence, SKSM 2017, Kuantan, Pahang, Malaysia, 27/8/17. https://doi.org/10.1063/1.5041544
Zamzuri ZH. The spatio-temporal multivariate Poisson lognormal model. In Proceeding of the 25th National Symposium on Mathematical Sciences, SKSM 2017: Mathematical Sciences as the Core of Intellectual Excellence. Vol. 1974. American Institute of Physics Inc. 2018. 020013 https://doi.org/10.1063/1.5041544
Zamzuri, Zamira Hasanah. / The spatio-temporal multivariate Poisson lognormal model. Proceeding of the 25th National Symposium on Mathematical Sciences, SKSM 2017: Mathematical Sciences as the Core of Intellectual Excellence. Vol. 1974 American Institute of Physics Inc., 2018.
@inproceedings{20c44bf7e6274a6d8f2f130d27479a98,
title = "The spatio-temporal multivariate Poisson lognormal model",
abstract = "To deal with the variation and correlation structure of accident data along with recognized covariate effects, we develop a spatio-temporal model for multivariate accident count data. Based on the multivariate Poisson lognormal model, we introduce linear combinations of random impulses to capture spatial correlation. For temporal effects, the lagged observations are added to this model. Model estimation is carried out using Markov Chain Monte Carlo methods. Simulated data sets are used in assessing the performance of this model. An advantage of this new model is that it not only copes with three sources of variations; time, space and multivariate data variations, but also provides information on time and space dependency. The model is also capable of providing an improvement in the accuracy of count data modelling.",
keywords = "count model, multivariate, traffic accident",
author = "Zamzuri, {Zamira Hasanah}",
year = "2018",
month = "6",
day = "28",
doi = "10.1063/1.5041544",
language = "English",
volume = "1974",
booktitle = "Proceeding of the 25th National Symposium on Mathematical Sciences, SKSM 2017",
publisher = "American Institute of Physics Inc.",

}

TY - GEN

T1 - The spatio-temporal multivariate Poisson lognormal model

AU - Zamzuri, Zamira Hasanah

PY - 2018/6/28

Y1 - 2018/6/28

N2 - To deal with the variation and correlation structure of accident data along with recognized covariate effects, we develop a spatio-temporal model for multivariate accident count data. Based on the multivariate Poisson lognormal model, we introduce linear combinations of random impulses to capture spatial correlation. For temporal effects, the lagged observations are added to this model. Model estimation is carried out using Markov Chain Monte Carlo methods. Simulated data sets are used in assessing the performance of this model. An advantage of this new model is that it not only copes with three sources of variations; time, space and multivariate data variations, but also provides information on time and space dependency. The model is also capable of providing an improvement in the accuracy of count data modelling.

AB - To deal with the variation and correlation structure of accident data along with recognized covariate effects, we develop a spatio-temporal model for multivariate accident count data. Based on the multivariate Poisson lognormal model, we introduce linear combinations of random impulses to capture spatial correlation. For temporal effects, the lagged observations are added to this model. Model estimation is carried out using Markov Chain Monte Carlo methods. Simulated data sets are used in assessing the performance of this model. An advantage of this new model is that it not only copes with three sources of variations; time, space and multivariate data variations, but also provides information on time and space dependency. The model is also capable of providing an improvement in the accuracy of count data modelling.

KW - count model

KW - multivariate

KW - traffic accident

UR - http://www.scopus.com/inward/record.url?scp=85049783110&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85049783110&partnerID=8YFLogxK

U2 - 10.1063/1.5041544

DO - 10.1063/1.5041544

M3 - Conference contribution

AN - SCOPUS:85049783110

VL - 1974

BT - Proceeding of the 25th National Symposium on Mathematical Sciences, SKSM 2017

PB - American Institute of Physics Inc.

ER -