Taking error into account when fitting models using Approximate Bayesian Computation

Elske van der Vaart, Dennis Prangle, Richard M. Sibly

Research output: Contribution to journalArticleScientificpeer-review

Abstract

Stochastic computer simulations are often the only practical way of answering questions relating to ecological management. However, due to their complexity, such models are difficult to calibrate and evaluate. Approximate Bayesian Computation (ABC) offers an increasingly popular approach to this problem, widely applied across a variety of fields. However, ensuring the accuracy of ABC's estimates has been difficult. Here, we obtain more accurate estimates by incorporating estimation of error into the ABC protocol. We show how this can be done where the data consist of repeated measures of the same quantity and errors may be assumed to be normally distributed and independent. We then derive the correct acceptance probabilities for a probabilistic ABC algorithm, and update the coverage test with which accuracy is assessed. We apply this method, which we call error-calibrated ABC, to a toy example and a realistic 14-parameter simulation model of earthworms that is used in environmental risk assessment. A comparison with exact methods and the diagnostic coverage test show that our approach improves estimation of parameter values and their credible intervals for both models.

Original languageEnglish
Pages (from-to)267-274
Number of pages8
JournalEcological applications
Volume28
Issue number2
DOIs
Publication statusPublished - Mar 2018

Keywords

  • Approximate Bayesian Computation (ABC)
  • IBM
  • individual-based model
  • model fitting
  • parameter estimation
  • stochastic computer simulation
  • EISENIA-FOETIDA SAVIGNY
  • INDIVIDUAL-BASED MODELS
  • MONTE-CARLO
  • POPULATION
  • CALIBRATION
  • GROWTH
  • INFERENCE
  • LIKELIHOODS
  • FECUNDITY
  • DYNAMICS

Cite this

van der Vaart, Elske ; Prangle, Dennis ; Sibly, Richard M. / Taking error into account when fitting models using Approximate Bayesian Computation. In: Ecological applications. 2018 ; Vol. 28, No. 2. pp. 267-274.
@article{226c01b493b34468afb528bd2edf7d66,
title = "Taking error into account when fitting models using Approximate Bayesian Computation",
abstract = "Stochastic computer simulations are often the only practical way of answering questions relating to ecological management. However, due to their complexity, such models are difficult to calibrate and evaluate. Approximate Bayesian Computation (ABC) offers an increasingly popular approach to this problem, widely applied across a variety of fields. However, ensuring the accuracy of ABC's estimates has been difficult. Here, we obtain more accurate estimates by incorporating estimation of error into the ABC protocol. We show how this can be done where the data consist of repeated measures of the same quantity and errors may be assumed to be normally distributed and independent. We then derive the correct acceptance probabilities for a probabilistic ABC algorithm, and update the coverage test with which accuracy is assessed. We apply this method, which we call error-calibrated ABC, to a toy example and a realistic 14-parameter simulation model of earthworms that is used in environmental risk assessment. A comparison with exact methods and the diagnostic coverage test show that our approach improves estimation of parameter values and their credible intervals for both models.",
keywords = "Approximate Bayesian Computation (ABC), IBM, individual-based model, model fitting, parameter estimation, stochastic computer simulation, EISENIA-FOETIDA SAVIGNY, INDIVIDUAL-BASED MODELS, MONTE-CARLO, POPULATION, CALIBRATION, GROWTH, INFERENCE, LIKELIHOODS, FECUNDITY, DYNAMICS",
author = "{van der Vaart}, Elske and Dennis Prangle and Sibly, {Richard M.}",
year = "2018",
month = "3",
doi = "10.1002/eap.1656",
language = "English",
volume = "28",
pages = "267--274",
journal = "Ecological applications",
issn = "1051-0761",
publisher = "Wiley",
number = "2",

}

Taking error into account when fitting models using Approximate Bayesian Computation. / van der Vaart, Elske; Prangle, Dennis; Sibly, Richard M.

In: Ecological applications, Vol. 28, No. 2, 03.2018, p. 267-274.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Taking error into account when fitting models using Approximate Bayesian Computation

AU - van der Vaart, Elske

AU - Prangle, Dennis

AU - Sibly, Richard M.

PY - 2018/3

Y1 - 2018/3

N2 - Stochastic computer simulations are often the only practical way of answering questions relating to ecological management. However, due to their complexity, such models are difficult to calibrate and evaluate. Approximate Bayesian Computation (ABC) offers an increasingly popular approach to this problem, widely applied across a variety of fields. However, ensuring the accuracy of ABC's estimates has been difficult. Here, we obtain more accurate estimates by incorporating estimation of error into the ABC protocol. We show how this can be done where the data consist of repeated measures of the same quantity and errors may be assumed to be normally distributed and independent. We then derive the correct acceptance probabilities for a probabilistic ABC algorithm, and update the coverage test with which accuracy is assessed. We apply this method, which we call error-calibrated ABC, to a toy example and a realistic 14-parameter simulation model of earthworms that is used in environmental risk assessment. A comparison with exact methods and the diagnostic coverage test show that our approach improves estimation of parameter values and their credible intervals for both models.

AB - Stochastic computer simulations are often the only practical way of answering questions relating to ecological management. However, due to their complexity, such models are difficult to calibrate and evaluate. Approximate Bayesian Computation (ABC) offers an increasingly popular approach to this problem, widely applied across a variety of fields. However, ensuring the accuracy of ABC's estimates has been difficult. Here, we obtain more accurate estimates by incorporating estimation of error into the ABC protocol. We show how this can be done where the data consist of repeated measures of the same quantity and errors may be assumed to be normally distributed and independent. We then derive the correct acceptance probabilities for a probabilistic ABC algorithm, and update the coverage test with which accuracy is assessed. We apply this method, which we call error-calibrated ABC, to a toy example and a realistic 14-parameter simulation model of earthworms that is used in environmental risk assessment. A comparison with exact methods and the diagnostic coverage test show that our approach improves estimation of parameter values and their credible intervals for both models.

KW - Approximate Bayesian Computation (ABC)

KW - IBM

KW - individual-based model

KW - model fitting

KW - parameter estimation

KW - stochastic computer simulation

KW - EISENIA-FOETIDA SAVIGNY

KW - INDIVIDUAL-BASED MODELS

KW - MONTE-CARLO

KW - POPULATION

KW - CALIBRATION

KW - GROWTH

KW - INFERENCE

KW - LIKELIHOODS

KW - FECUNDITY

KW - DYNAMICS

U2 - 10.1002/eap.1656

DO - 10.1002/eap.1656

M3 - Article

VL - 28

SP - 267

EP - 274

JO - Ecological applications

JF - Ecological applications

SN - 1051-0761

IS - 2

ER -