A comparison of likelihood-based spatiotemporal monitoring methods under non-homogenous population size

Sung Won Han, Wei Jiang, Lianjie Shu, Kwok Leung Tsui

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

This article discusses the spatiotemporal surveillance problem of detecting rate changes of Poisson data considering non-homogenous population sample size. By applying Monte Carlo simulations, we investigate the performance of several likelihood-based approaches under various scenarios depending on four factors: (1) population trend, (2) change magnitude, (3) change coverage, and (4) change time. Our article evaluates the performance of spatiotemporal surveillance methods based on the average run length at different change times. The simulation results show that no method is uniformly better than others in all scenarios. The difference between the generalized likelihood ratio (GLR) approach and the weighted likelihood ratio (WLR) approach depends mainly on population size, not change coverage, change magnitude, or change time. We find that changes associated with a small population in time periods and/or spatial regions favor the WLR approach, but those associated with a large population favor the GLR under any trends of population changes.

Original languageEnglish
Pages (from-to)14-39
Number of pages26
JournalCommunications in Statistics: Simulation and Computation
Volume44
Issue number1
DOIs
Publication statusPublished - 2015 Jan 2
Externally publishedYes

Fingerprint

Likelihood Ratio
Population Size
Time Change
Likelihood
Monitoring
Weighted Likelihood
Surveillance
Coverage
Average Run Length
Scenarios
Rate of change
Siméon Denis Poisson
Sample Size
Monte Carlo Simulation
Monte Carlo simulation
Evaluate
Simulation
Trends

Keywords

  • Change point detection
  • Generalized likelihood ratio
  • Non-homogenous Poisson
  • Scan statistics
  • Spatiotemporal surveillance
  • Weighted likelihood ratio

ASJC Scopus subject areas

  • Statistics and Probability
  • Modelling and Simulation

Cite this

A comparison of likelihood-based spatiotemporal monitoring methods under non-homogenous population size. / Han, Sung Won; Jiang, Wei; Shu, Lianjie; Tsui, Kwok Leung.

In: Communications in Statistics: Simulation and Computation, Vol. 44, No. 1, 02.01.2015, p. 14-39.

Research output: Contribution to journalArticle

@article{8c469e6e3b0c40fb8961125015e916aa,
title = "A comparison of likelihood-based spatiotemporal monitoring methods under non-homogenous population size",
abstract = "This article discusses the spatiotemporal surveillance problem of detecting rate changes of Poisson data considering non-homogenous population sample size. By applying Monte Carlo simulations, we investigate the performance of several likelihood-based approaches under various scenarios depending on four factors: (1) population trend, (2) change magnitude, (3) change coverage, and (4) change time. Our article evaluates the performance of spatiotemporal surveillance methods based on the average run length at different change times. The simulation results show that no method is uniformly better than others in all scenarios. The difference between the generalized likelihood ratio (GLR) approach and the weighted likelihood ratio (WLR) approach depends mainly on population size, not change coverage, change magnitude, or change time. We find that changes associated with a small population in time periods and/or spatial regions favor the WLR approach, but those associated with a large population favor the GLR under any trends of population changes.",
keywords = "Change point detection, Generalized likelihood ratio, Non-homogenous Poisson, Scan statistics, Spatiotemporal surveillance, Weighted likelihood ratio",
author = "Han, {Sung Won} and Wei Jiang and Lianjie Shu and Tsui, {Kwok Leung}",
year = "2015",
month = "1",
day = "2",
doi = "10.1080/03610918.2013.763981",
language = "English",
volume = "44",
pages = "14--39",
journal = "Communications in Statistics Part B: Simulation and Computation",
issn = "0361-0918",
publisher = "Taylor and Francis Ltd.",
number = "1",

}

TY - JOUR

T1 - A comparison of likelihood-based spatiotemporal monitoring methods under non-homogenous population size

AU - Han, Sung Won

AU - Jiang, Wei

AU - Shu, Lianjie

AU - Tsui, Kwok Leung

PY - 2015/1/2

Y1 - 2015/1/2

N2 - This article discusses the spatiotemporal surveillance problem of detecting rate changes of Poisson data considering non-homogenous population sample size. By applying Monte Carlo simulations, we investigate the performance of several likelihood-based approaches under various scenarios depending on four factors: (1) population trend, (2) change magnitude, (3) change coverage, and (4) change time. Our article evaluates the performance of spatiotemporal surveillance methods based on the average run length at different change times. The simulation results show that no method is uniformly better than others in all scenarios. The difference between the generalized likelihood ratio (GLR) approach and the weighted likelihood ratio (WLR) approach depends mainly on population size, not change coverage, change magnitude, or change time. We find that changes associated with a small population in time periods and/or spatial regions favor the WLR approach, but those associated with a large population favor the GLR under any trends of population changes.

AB - This article discusses the spatiotemporal surveillance problem of detecting rate changes of Poisson data considering non-homogenous population sample size. By applying Monte Carlo simulations, we investigate the performance of several likelihood-based approaches under various scenarios depending on four factors: (1) population trend, (2) change magnitude, (3) change coverage, and (4) change time. Our article evaluates the performance of spatiotemporal surveillance methods based on the average run length at different change times. The simulation results show that no method is uniformly better than others in all scenarios. The difference between the generalized likelihood ratio (GLR) approach and the weighted likelihood ratio (WLR) approach depends mainly on population size, not change coverage, change magnitude, or change time. We find that changes associated with a small population in time periods and/or spatial regions favor the WLR approach, but those associated with a large population favor the GLR under any trends of population changes.

KW - Change point detection

KW - Generalized likelihood ratio

KW - Non-homogenous Poisson

KW - Scan statistics

KW - Spatiotemporal surveillance

KW - Weighted likelihood ratio

UR - http://www.scopus.com/inward/record.url?scp=84905483312&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84905483312&partnerID=8YFLogxK

U2 - 10.1080/03610918.2013.763981

DO - 10.1080/03610918.2013.763981

M3 - Article

VL - 44

SP - 14

EP - 39

JO - Communications in Statistics Part B: Simulation and Computation

JF - Communications in Statistics Part B: Simulation and Computation

SN - 0361-0918

IS - 1

ER -