Difference between revisions of "Bayes Factor"
m |
|||
| Line 5: | Line 5: | ||
| − | showing that it may be calculated directly using the [[joint probability distribution]] or from the [[conditional probability distribution]] where the second form of the integral was obtained using the [[Bayesian Product Rule]]. | + | showing that it may be calculated directly using the [[joint probability distribution]] or from the [[conditional probability distribution]] where the second form of the integral was obtained using the [[Bayesian Product Rule]]. Thus the ''Bayes factor'' for model class M<sub>1</sub> with parameters <math>\theta_1</math> to model class M<sub>2</sub> with parameters <math>\theta_2</math> is |
| + | |||
| + | ::<math>BF(X,M_1,M_2) = \frac{m_{\theta_1}(X|M_1)}{m_{\theta_2}(X|M_2)} </math> | ||
| + | |||
| + | |||
| + | In determining which model class has the largest ratio when compared against all others, the model class or [[mathematical model]] which best explains the data is determined. It is then left to determine the best inference as to the values of that models parameters via [[parameter estimation]]. | ||
[[Category:Probability]] | [[Category:Probability]] | ||
Revision as of 13:31, December 28, 2007
The Bayes Factor of model class M1 to model class M2 for some set of observations X is the ratio of their associated marginal class likelihoods. The marginal class likelihood of some set of observations is the marginal likelihood of the observations for the model class obtained by marginalizing the joint probability distribution of the observations and the model class parameters by treating the model class parameters Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \theta} as nuisance parameters, i.e.,
- Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle m_{\theta}(X|M) = \int_\theta p(X,\theta|M) \, d\theta = \int_\theta p(X|\theta , M) \, p(\theta|M) \, d\theta }
showing that it may be calculated directly using the joint probability distribution or from the conditional probability distribution where the second form of the integral was obtained using the Bayesian Product Rule. Thus the Bayes factor for model class M1 with parameters Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \theta_1}
to model class M2 with parameters Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \theta_2}
is
- Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle BF(X,M_1,M_2) = \frac{m_{\theta_1}(X|M_1)}{m_{\theta_2}(X|M_2)} }
In determining which model class has the largest ratio when compared against all others, the model class or mathematical model which best explains the data is determined. It is then left to determine the best inference as to the values of that models parameters via parameter estimation.