Skip to content
Back to outputs

Galactic double neutron star total masses and Gaussian mixture model selection

Research output: Contribution to journalArticle

Standard

Galactic double neutron star total masses and Gaussian mixture model selection. / Keitel, David.

In: Monthly Notices of the Royal Astronomical Society, Vol. 485, No. 2, stz358, 11.05.2019, p. 1665-1674.

Research output: Contribution to journalArticle

Harvard

Keitel, D 2019, 'Galactic double neutron star total masses and Gaussian mixture model selection', Monthly Notices of the Royal Astronomical Society, vol. 485, no. 2, stz358, pp. 1665-1674. https://doi.org/10.1093/mnras/stz358

APA

Keitel, D. (2019). Galactic double neutron star total masses and Gaussian mixture model selection. Monthly Notices of the Royal Astronomical Society, 485(2), 1665-1674. [stz358]. https://doi.org/10.1093/mnras/stz358

Vancouver

Keitel D. Galactic double neutron star total masses and Gaussian mixture model selection. Monthly Notices of the Royal Astronomical Society. 2019 May 11;485(2):1665-1674. stz358. https://doi.org/10.1093/mnras/stz358

Author

Keitel, David. / Galactic double neutron star total masses and Gaussian mixture model selection. In: Monthly Notices of the Royal Astronomical Society. 2019 ; Vol. 485, No. 2. pp. 1665-1674.

Bibtex

@article{4b5055dabc364c048f2c82c525ef4e7a,
title = "Galactic double neutron star total masses and Gaussian mixture model selection",
abstract = "Huang et al. (2018) have analysed the population of 15 known Galactic double neutron stars (DNSs) regarding the total masses of these systems. They suggest the existence of two subpopulations, and report likelihood-based preference for a two-component Gaussian mixture model over a single-Gaussian distribution. This note offers a cautionary perspective on model selection for this data set: especially for such a small sample size, a pure likelihood ratio test can encourage overfitting. This can be avoided by penalizing models with a higher number of free parameters. Re-examining the DNS total mass data set within the class of Gaussian mixture models, this can be achieved through several simple and well-established statistical tests, including information criteria (AICc, BIC), cross-validation, Bayesian evidence ratios, and a penalized EM-test. While this reanalysis confirms the basic finding that a two-component mixture is consistent with the data, the model selection criteria consistently indicate that there is no robust preference for it over a single-component fit. Additional DNS discoveries will be needed to settle the question of subpopulations.",
keywords = "astro-ph.HE",
author = "David Keitel",
year = "2019",
month = may
day = "11",
doi = "10.1093/mnras/stz358",
language = "English",
volume = "485",
pages = "1665--1674",
journal = "MNRAS",
issn = "0035-8711",
publisher = "Oxford University Press",
number = "2",

}

RIS

TY - JOUR

T1 - Galactic double neutron star total masses and Gaussian mixture model selection

AU - Keitel, David

PY - 2019/5/11

Y1 - 2019/5/11

N2 - Huang et al. (2018) have analysed the population of 15 known Galactic double neutron stars (DNSs) regarding the total masses of these systems. They suggest the existence of two subpopulations, and report likelihood-based preference for a two-component Gaussian mixture model over a single-Gaussian distribution. This note offers a cautionary perspective on model selection for this data set: especially for such a small sample size, a pure likelihood ratio test can encourage overfitting. This can be avoided by penalizing models with a higher number of free parameters. Re-examining the DNS total mass data set within the class of Gaussian mixture models, this can be achieved through several simple and well-established statistical tests, including information criteria (AICc, BIC), cross-validation, Bayesian evidence ratios, and a penalized EM-test. While this reanalysis confirms the basic finding that a two-component mixture is consistent with the data, the model selection criteria consistently indicate that there is no robust preference for it over a single-component fit. Additional DNS discoveries will be needed to settle the question of subpopulations.

AB - Huang et al. (2018) have analysed the population of 15 known Galactic double neutron stars (DNSs) regarding the total masses of these systems. They suggest the existence of two subpopulations, and report likelihood-based preference for a two-component Gaussian mixture model over a single-Gaussian distribution. This note offers a cautionary perspective on model selection for this data set: especially for such a small sample size, a pure likelihood ratio test can encourage overfitting. This can be avoided by penalizing models with a higher number of free parameters. Re-examining the DNS total mass data set within the class of Gaussian mixture models, this can be achieved through several simple and well-established statistical tests, including information criteria (AICc, BIC), cross-validation, Bayesian evidence ratios, and a penalized EM-test. While this reanalysis confirms the basic finding that a two-component mixture is consistent with the data, the model selection criteria consistently indicate that there is no robust preference for it over a single-component fit. Additional DNS discoveries will be needed to settle the question of subpopulations.

KW - astro-ph.HE

U2 - 10.1093/mnras/stz358

DO - 10.1093/mnras/stz358

M3 - Article

VL - 485

SP - 1665

EP - 1674

JO - MNRAS

JF - MNRAS

SN - 0035-8711

IS - 2

M1 - stz358

ER -

ID: 12518104