1988
DOI: 10.1214/aos/1176350709
|View full text |Cite
|
Sign up to set email alerts
|

On the Choice of a Model to Fit Data from an Exponential Family

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

1
131
0

Year Published

1997
1997
2016
2016

Publication Types

Select...
5
3
1

Relationship

0
9

Authors

Journals

citations
Cited by 238 publications
(132 citation statements)
references
References 16 publications
1
131
0
Order By: Relevance
“…This theorem is proven in [56,Thm. 2.3], where a more refined expansion gives a remainder that is bounded in probability when multiplied by √ n. Theorem 5.1.5 shows that model selection using the BIC approximates a Bayesian procedure seeking the model with highest posterior probability.…”
Section: Information Criteria and Asymptoticsmentioning
confidence: 78%
See 1 more Smart Citation
“…This theorem is proven in [56,Thm. 2.3], where a more refined expansion gives a remainder that is bounded in probability when multiplied by √ n. Theorem 5.1.5 shows that model selection using the BIC approximates a Bayesian procedure seeking the model with highest posterior probability.…”
Section: Information Criteria and Asymptoticsmentioning
confidence: 78%
“…1.2]. Note that in [56] the result is stated for the case where Θ 1 and Θ 2 are smooth manifolds but this property is not used in the proof. In algebraic statistics, Θ 1 and Θ 2 will be semi-algebraic.…”
Section: Information Criteria and Asymptoticsmentioning
confidence: 99%
“…The model with smaller Bayesian Information Criterion (BIC) or Schartz Information Criterion was chosen among all the possible models formed with combinations of the five SNP markers (Haughton, 1988). Similar results were obtained when using the Akaike Information Criterion (Burnham and Anderson, 2002).…”
Section: A Koets Et Al / Preventive Veterinary Medicine XXX (2009) mentioning
confidence: 85%
“…was inferred by finding the graphical model that minimises the Bayesian information criterion as implemented in the R package gRapHD (Abreu et al, 2010). This inference procedure has optimal properties (see Haughton, 1988). Confirmatory analyses were performed using a Monte Carlo version of the Fisher's exact test with 1 000 000 Monte Carlo samples.…”
Section: Discussionmentioning
confidence: 99%