A Bayesian Nonparametric Spiked Process Prior for Dynamic Model Selection

Alberto Cassese, Weixuan Zhu, Michele Guindani, Marina Vannucci

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

In many applications, investigators monitor processes that vary in space and time, with the goal of identifying temporally persistent and spatially localized departures from a baseline or "normal" behavior. In this manuscript, we consider the monitoring of pneumonia and influenza (P&I) mortality, to detect influenza outbreaks in the continental United States, and propose a Bayesian nonparametric model selection approach to take into account the spatio-temporal dependence of outbreaks. More specifically, we introduce a zero-inflated conditionally identically distributed species sampling prior which allows borrowing information across time and to assign data to clusters associated to either a null or an alternate process. Spatial dependences are accounted for by means of a Markov random field prior, which allows to inform the selection based on inferences conducted at nearby locations. We show how the proposed modeling framework performs in an application to the P&I mortality data and in a simulation study, and compare with common threshold methods for detecting outbreaks over time, with more recent Markov switching based models, and with spike-and-slab Bayesian nonparametric priors that do not take into account spatio-temporal dependence.

Original languageEnglish
Pages (from-to)553-572
Number of pages20
JournalBayesian Analysis
Volume14
Issue number2
Early online dateAug 2018
DOIs
Publication statusPublished - 2019

Keywords

  • nonparametric Bayes
  • variable selection
  • Markov random field
  • spatio-temporal data
  • VARIABLE SELECTION
  • INFLUENZA

Cite this

Cassese, Alberto ; Zhu, Weixuan ; Guindani, Michele ; Vannucci, Marina. / A Bayesian Nonparametric Spiked Process Prior for Dynamic Model Selection. In: Bayesian Analysis. 2019 ; Vol. 14, No. 2. pp. 553-572.
@article{80ff63f76bd3421a83cf17d273599e5d,
title = "A Bayesian Nonparametric Spiked Process Prior for Dynamic Model Selection",
abstract = "In many applications, investigators monitor processes that vary in space and time, with the goal of identifying temporally persistent and spatially localized departures from a baseline or {"}normal{"} behavior. In this manuscript, we consider the monitoring of pneumonia and influenza (P&I) mortality, to detect influenza outbreaks in the continental United States, and propose a Bayesian nonparametric model selection approach to take into account the spatio-temporal dependence of outbreaks. More specifically, we introduce a zero-inflated conditionally identically distributed species sampling prior which allows borrowing information across time and to assign data to clusters associated to either a null or an alternate process. Spatial dependences are accounted for by means of a Markov random field prior, which allows to inform the selection based on inferences conducted at nearby locations. We show how the proposed modeling framework performs in an application to the P&I mortality data and in a simulation study, and compare with common threshold methods for detecting outbreaks over time, with more recent Markov switching based models, and with spike-and-slab Bayesian nonparametric priors that do not take into account spatio-temporal dependence.",
keywords = "nonparametric Bayes, variable selection, Markov random field, spatio-temporal data, VARIABLE SELECTION, INFLUENZA",
author = "Alberto Cassese and Weixuan Zhu and Michele Guindani and Marina Vannucci",
year = "2019",
doi = "10.1214/18-BA1116",
language = "English",
volume = "14",
pages = "553--572",
journal = "Bayesian Analysis",
issn = "1931-6690",
publisher = "Carnegie Mellon University",
number = "2",

}

A Bayesian Nonparametric Spiked Process Prior for Dynamic Model Selection. / Cassese, Alberto; Zhu, Weixuan; Guindani, Michele; Vannucci, Marina.

In: Bayesian Analysis, Vol. 14, No. 2, 2019, p. 553-572.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - A Bayesian Nonparametric Spiked Process Prior for Dynamic Model Selection

AU - Cassese, Alberto

AU - Zhu, Weixuan

AU - Guindani, Michele

AU - Vannucci, Marina

PY - 2019

Y1 - 2019

N2 - In many applications, investigators monitor processes that vary in space and time, with the goal of identifying temporally persistent and spatially localized departures from a baseline or "normal" behavior. In this manuscript, we consider the monitoring of pneumonia and influenza (P&I) mortality, to detect influenza outbreaks in the continental United States, and propose a Bayesian nonparametric model selection approach to take into account the spatio-temporal dependence of outbreaks. More specifically, we introduce a zero-inflated conditionally identically distributed species sampling prior which allows borrowing information across time and to assign data to clusters associated to either a null or an alternate process. Spatial dependences are accounted for by means of a Markov random field prior, which allows to inform the selection based on inferences conducted at nearby locations. We show how the proposed modeling framework performs in an application to the P&I mortality data and in a simulation study, and compare with common threshold methods for detecting outbreaks over time, with more recent Markov switching based models, and with spike-and-slab Bayesian nonparametric priors that do not take into account spatio-temporal dependence.

AB - In many applications, investigators monitor processes that vary in space and time, with the goal of identifying temporally persistent and spatially localized departures from a baseline or "normal" behavior. In this manuscript, we consider the monitoring of pneumonia and influenza (P&I) mortality, to detect influenza outbreaks in the continental United States, and propose a Bayesian nonparametric model selection approach to take into account the spatio-temporal dependence of outbreaks. More specifically, we introduce a zero-inflated conditionally identically distributed species sampling prior which allows borrowing information across time and to assign data to clusters associated to either a null or an alternate process. Spatial dependences are accounted for by means of a Markov random field prior, which allows to inform the selection based on inferences conducted at nearby locations. We show how the proposed modeling framework performs in an application to the P&I mortality data and in a simulation study, and compare with common threshold methods for detecting outbreaks over time, with more recent Markov switching based models, and with spike-and-slab Bayesian nonparametric priors that do not take into account spatio-temporal dependence.

KW - nonparametric Bayes

KW - variable selection

KW - Markov random field

KW - spatio-temporal data

KW - VARIABLE SELECTION

KW - INFLUENZA

U2 - 10.1214/18-BA1116

DO - 10.1214/18-BA1116

M3 - Article

VL - 14

SP - 553

EP - 572

JO - Bayesian Analysis

JF - Bayesian Analysis

SN - 1931-6690

IS - 2

ER -