Bayes factor

Last updated

The Bayes factor is a ratio of two competing statistical models represented by their evidence, and is used to quantify the support for one model over the other. [1] The models in questions can have a common set of parameters, such as a null hypothesis and an alternative, but this is not necessary; for instance, it could also be a non-linear model compared to its linear approximation. The Bayes factor can be thought of as a Bayesian analog to the likelihood-ratio test, although it uses the integrated (i.e., marginal) likelihood rather than the maximized likelihood. As such, both quantities only coincide under simple hypotheses (e.g., two specific parameter values). [2] Also, in contrast with null hypothesis significance testing, Bayes factors support evaluation of evidence in favor of a null hypothesis, rather than only allowing the null to be rejected or not rejected. [3]

Contents

Although conceptually simple, the computation of the Bayes factor can be challenging depending on the complexity of the model and the hypotheses. [4] Since closed-form expressions of the marginal likelihood are generally not available, numerical approximations based on MCMC samples have been suggested. [5] For certain special cases, simplified algebraic expressions can be derived; for instance, the Savage–Dickey density ratio in the case of a precise (equality constrained) hypothesis against an unrestricted alternative. [6] [7] Another approximation, derived by applying Laplace's approximation to the integrated likelihoods, is known as the Bayesian information criterion (BIC); [8] in large data sets the Bayes factor will approach the BIC as the influence of the priors wanes. In small data sets, priors generally matter and must not be improper since the Bayes factor will be undefined if either of the two integrals in its ratio is not finite.

Definition

The Bayes factor is the ratio of two marginal likelihoods; that is, the likelihoods of two statistical models integrated over the prior probabilities of their parameters. [9]

The posterior probability of a model M given data D is given by Bayes' theorem:

The key data-dependent term represents the probability that some data are produced under the assumption of the model M; evaluating it correctly is the key to Bayesian model comparison.

Given a model selection problem in which one wishes to choose between two models on the basis of observed data D, the plausibility of the two different models M1 and M2, parametrised by model parameter vectors and , is assessed by the Bayes factor K given by

When the two models have equal prior probability, so that , the Bayes factor is equal to the ratio of the posterior probabilities of M1 and M2. If instead of the Bayes factor integral, the likelihood corresponding to the maximum likelihood estimate of the parameter for each statistical model is used, then the test becomes a classical likelihood-ratio test. Unlike a likelihood-ratio test, this Bayesian model comparison does not depend on any single set of parameters, as it integrates over all parameters in each model (with respect to the respective priors). An advantage of the use of Bayes factors is that it automatically, and quite naturally, includes a penalty for including too much model structure. [10] It thus guards against overfitting. For models where an explicit version of the likelihood is not available or too costly to evaluate numerically, approximate Bayesian computation can be used for model selection in a Bayesian framework, [11] with the caveat that approximate-Bayesian estimates of Bayes factors are often biased. [12]

Other approaches are:

Interpretation

A value of K > 1 means that M1 is more strongly supported by the data under consideration than M2. Note that classical hypothesis testing gives one hypothesis (or model) preferred status (the 'null hypothesis'), and only considers evidence against it. Harold Jeffreys gave a scale for interpretation of K: [13]

KdHartbitsStrength of evidence
< 100< 0< 0Negative (supports M2)
100 to 101/20 to 50 to 1.6Barely worth mentioning
101/2 to 1015 to 101.6 to 3.3Substantial
101 to 103/210 to 153.3 to 5.0Strong
103/2 to 10215 to 205.0 to 6.6Very strong
> 102> 20> 6.6Decisive

The second column gives the corresponding weights of evidence in decihartleys (also known as decibans); bits are added in the third column for clarity. According to I. J. Good a change in a weight of evidence of 1 deciban or 1/3 of a bit (i.e. a change in an odds ratio from evens to about 5:4) is about as finely as humans can reasonably perceive their degree of belief in a hypothesis in everyday use. [14]

An alternative table, widely cited, is provided by Kass and Raftery (1995): [10]

log10KKStrength of evidence
0 to 1/21 to 3.2Not worth more than a bare mention
1/2 to 13.2 to 10Substantial
1 to 210 to 100Strong
> 2> 100Decisive

Example

Suppose we have a random variable that produces either a success or a failure. We want to compare a model M1 where the probability of success is q = 12, and another model M2 where q is unknown and we take a prior distribution for q that is uniform on [0,1]. We take a sample of 200, and find 115 successes and 85 failures. The likelihood can be calculated according to the binomial distribution:

Thus we have for M1

whereas for M2 we have

The ratio is then 1.2, which is "barely worth mentioning" even if it points very slightly towards M1.

A frequentist hypothesis test of M1 (here considered as a null hypothesis) would have produced a very different result. Such a test says that M1 should be rejected at the 5% significance level, since the probability of getting 115 or more successes from a sample of 200 if q = 12 is 0.02, and as a two-tailed test of getting a figure as extreme as or more extreme than 115 is 0.04. Note that 115 is more than two standard deviations away from 100. Thus, whereas a frequentist hypothesis test would yield significant results at the 5% significance level, the Bayes factor hardly considers this to be an extreme result. Note, however, that a non-uniform prior (for example one that reflects the fact that you expect the number of success and failures to be of the same order of magnitude) could result in a Bayes factor that is more in agreement with the frequentist hypothesis test.

A classical likelihood-ratio test would have found the maximum likelihood estimate for q, namely , whence

(rather than averaging over all possible q). That gives a likelihood ratio of 0.1 and points towards M2.

M2 is a more complex model than M1 because it has a free parameter which allows it to model the data more closely. The ability of Bayes factors to take this into account is a reason why Bayesian inference has been put forward as a theoretical justification for and generalisation of Occam's razor, reducing Type I errors. [15]

On the other hand, the modern method of relative likelihood takes into account the number of free parameters in the models, unlike the classical likelihood ratio. The relative likelihood method could be applied as follows. Model M1 has 0 parameters, and so its Akaike information criterion (AIC) value is . Model M2 has 1 parameter, and so its AIC value is . Hence M1 is about times as probable as M2 to minimize the information loss. Thus M2 is slightly preferred, but M1 cannot be excluded.

See also

Statistical ratios

Related Research Articles

The likelihood function is the joint probability mass of observed data viewed as a function of the parameters of a statistical model. Intuitively, the likelihood function is the probability of observing data assuming is the actual parameter.

In statistics, the likelihood-ratio test assesses the goodness of fit of two competing statistical models, specifically one found by maximization over the entire parameter space and another found after imposing some constraint, based on the ratio of their likelihoods. If the constraint is supported by the observed data, the two likelihoods should not differ by more than sampling error. Thus the likelihood-ratio test tests whether this ratio is significantly different from one, or equivalently whether its natural logarithm is significantly different from zero.

Bayesian inference is a method of statistical inference in which Bayes' theorem is used to update the probability for a hypothesis as more evidence or information becomes available. Fundamentally, Bayesian inference uses prior knowledge, in the form of a prior distribution in order to estimate posterior probabilities. Bayesian inference is an important technique in statistics, and especially in mathematical statistics. Bayesian updating is particularly important in the dynamic analysis of a sequence of data. Bayesian inference has found application in a wide range of activities, including science, engineering, philosophy, medicine, sport, and law. In the philosophy of decision theory, Bayesian inference is closely related to subjective probability, often called "Bayesian probability".

In statistics, maximum likelihood estimation (MLE) is a method of estimating the parameters of an assumed probability distribution, given some observed data. This is achieved by maximizing a likelihood function so that, under the assumed statistical model, the observed data is most probable. The point in the parameter space that maximizes the likelihood function is called the maximum likelihood estimate. The logic of maximum likelihood is both intuitive and flexible, and as such the method has become a dominant means of statistical inference.

A Bayesian network is a probabilistic graphical model that represents a set of variables and their conditional dependencies via a directed acyclic graph (DAG). While it is one of several forms of causal notation, causal networks are special cases of Bayesian networks. Bayesian networks are ideal for taking an event that occurred and predicting the likelihood that any one of several possible known causes was the contributing factor. For example, a Bayesian network could represent the probabilistic relationships between diseases and symptoms. Given symptoms, the network can be used to compute the probabilities of the presence of various diseases.

The posterior probability is a type of conditional probability that results from updating the prior probability with information summarized by the likelihood via an application of Bayes' rule. From an epistemological perspective, the posterior probability contains everything there is to know about an uncertain proposition, given prior knowledge and a mathematical model describing the observations available at a particular time. After the arrival of new information, the current posterior probability may serve as the prior in another round of Bayesian updating.

In statistics, the score is the gradient of the log-likelihood function with respect to the parameter vector. Evaluated at a particular point of the parameter vector, the score indicates the steepness of the log-likelihood function and thereby the sensitivity to infinitesimal changes to the parameter values. If the log-likelihood function is continuous over the parameter space, the score will vanish at a local maximum or minimum; this fact is used in maximum likelihood estimation to find the parameter values that maximize the likelihood function.

A marginal likelihood is a likelihood function that has been integrated over the parameter space. In Bayesian statistics, it represents the probability of generating the observed sample for all possible values of the parameters; it can be understood as the probability of the model itself and is therefore often referred to as model evidence or simply evidence.

In statistics, the score test assesses constraints on statistical parameters based on the gradient of the likelihood function—known as the score—evaluated at the hypothesized parameter value under the null hypothesis. Intuitively, if the restricted estimator is near the maximum of the likelihood function, the score should not differ from zero by more than sampling error. While the finite sample distributions of score tests are generally unknown, they have an asymptotic χ2-distribution under the null hypothesis as first proved by C. R. Rao in 1948, a fact that can be used to determine statistical significance.

In statistics, the Wald test assesses constraints on statistical parameters based on the weighted distance between the unrestricted estimate and its hypothesized value under the null hypothesis, where the weight is the precision of the estimate. Intuitively, the larger this weighted distance, the less likely it is that the constraint is true. While the finite sample distributions of Wald tests are generally unknown, it has an asymptotic χ2-distribution under the null hypothesis, a fact that can be used to determine statistical significance.

In Bayesian probability, the Jeffreys prior, named after Sir Harold Jeffreys, is a non-informative prior distribution for a parameter space; its density function is proportional to the square root of the determinant of the Fisher information matrix:

In statistics, the Bayesian information criterion (BIC) or Schwarz information criterion is a criterion for model selection among a finite set of models; models with lower BIC are generally preferred. It is based, in part, on the likelihood function and it is closely related to the Akaike information criterion (AIC).

<span class="mw-page-title-main">Monotone likelihood ratio</span> Statistical property

In statistics, the monotone likelihood ratio property is a property of the ratio of two probability density functions (PDFs). Formally, distributions ƒ(x) and g(x) bear the property if

Approximate Bayesian computation (ABC) constitutes a class of computational methods rooted in Bayesian statistics that can be used to estimate the posterior distributions of model parameters.

One-shot learning is an object categorization problem, found mostly in computer vision. Whereas most machine learning-based object categorization algorithms require training on hundreds or thousands of examples, one-shot learning aims to classify objects from one, or only a few, examples. The term few-shot learning is also used for these problems, especially when more than one example is needed.

Bayesian econometrics is a branch of econometrics which applies Bayesian principles to economic modelling. Bayesianism is based on a degree-of-belief interpretation of probability, as opposed to a relative-frequency interpretation.

In statistics Wilks' theorem states that the log-likelihood ratio is asymptotically normal. This can be used to produce confidence intervals for maximum-likelihood estimates or as a test statistic for performing the likelihood-ratio test.

In particle physics, CLs represents a statistical method for setting upper limits on model parameters, a particular form of interval estimation used for parameters that can take only non-negative values. Although CLs are said to refer to Confidence Levels, "The method's name is ... misleading, as the CLs exclusion region is not a confidence interval." It was first introduced by physicists working at the LEP experiment at CERN and has since been used by many high energy physics experiments. It is a frequentist method in the sense that the properties of the limit are defined by means of error probabilities, however it differs from standard confidence intervals in that the stated confidence level of the interval is not equal to its coverage probability. The reason for this deviation is that standard upper limits based on a most powerful test necessarily produce empty intervals with some fixed probability when the parameter value is zero, and this property is considered undesirable by most physicists and statisticians.

In statistics, when selecting a statistical model for given data, the relative likelihood compares the relative plausibilities of different candidate models or of different values of a parameter of a single model.

Bayesian model reduction is a method for computing the evidence and posterior over the parameters of Bayesian models that differ in their priors. A full model is fitted to data using standard approaches. Hypotheses are then tested by defining one or more 'reduced' models with alternative priors, which usually – in the limit – switch off certain parameters. The evidence and parameters of the reduced models can then be computed from the evidence and estimated (posterior) parameters of the full model using Bayesian model reduction. If the priors and posteriors are normally distributed, then there is an analytic solution which can be computed rapidly. This has multiple scientific and engineering applications: these include scoring the evidence for large numbers of models very quickly and facilitating the estimation of hierarchical models.

References

  1. Morey, Richard D.; Romeijn, Jan-Willem; Rouder, Jeffrey N. (2016). "The philosophy of Bayes factors and the quantification of statistical evidence". Journal of Mathematical Psychology. 72: 6–18. doi: 10.1016/j.jmp.2015.11.001 .
  2. Lesaffre, Emmanuel; Lawson, Andrew B. (2012). "Bayesian hypothesis testing". Bayesian Biostatistics. Somerset: John Wiley & Sons. pp. 72–78. doi:10.1002/9781119942412.ch3. ISBN   978-0-470-01823-1.
  3. Ly, Alexander; et al. (2020). "The Bayesian Methodology of Sir Harold Jeffreys as a Practical Alternative to the P Value Hypothesis Test". Computational Brain & Behavior. 3 (2): 153–161. doi: 10.1007/s42113-019-00070-x . hdl: 2066/226717 .
  4. Llorente, Fernando; et al. (2023). "Marginal likelihood computation for model selection and hypothesis testing: an extensive review". SIAM Review. to appear: 3–58. arXiv: 2005.08334 . doi:10.1137/20M1310849. S2CID   210156537.
  5. Congdon, Peter (2014). "Estimating model probabilities or marginal likelihoods in practice". Applied Bayesian Modelling (2nd ed.). Wiley. pp. 38–40. ISBN   978-1-119-95151-3.
  6. Koop, Gary (2003). "Model Comparison: The Savage–Dickey Density Ratio". Bayesian Econometrics. Somerset: John Wiley & Sons. pp. 69–71. ISBN   0-470-84567-8.
  7. Wagenmakers, Eric-Jan; Lodewyckx, Tom; Kuriyal, Himanshu; Grasman, Raoul (2010). "Bayesian hypothesis testing for psychologists: A tutorial on the Savage–Dickey method" (PDF). Cognitive Psychology. 60 (3): 158–189. doi:10.1016/j.cogpsych.2009.12.001. PMID   20064637. S2CID   206867662.
  8. Ibrahim, Joseph G.; Chen, Ming-Hui; Sinha, Debajyoti (2001). "Model Comparison". Bayesian Survival Analysis. Springer Series in Statistics. New York: Springer. pp. 246–254. doi:10.1007/978-1-4757-3447-8_6. ISBN   0-387-95277-2.
  9. Gill, Jeff (2002). "Bayesian Hypothesis Testing and the Bayes Factor". Bayesian Methods : A Social and Behavioral Sciences Approach. Chapman & Hall. pp. 199–237. ISBN   1-58488-288-3.
  10. 1 2 Robert E. Kass & Adrian E. Raftery (1995). "Bayes Factors" (PDF). Journal of the American Statistical Association. 90 (430): 791. doi:10.2307/2291091. JSTOR   2291091.
  11. Toni, T.; Stumpf, M.P.H. (2009). "Simulation-based model selection for dynamical systems in systems and population biology". Bioinformatics. 26 (1): 104–10. arXiv: 0911.1705 . doi:10.1093/bioinformatics/btp619. PMC   2796821 . PMID   19880371.
  12. Robert, C.P.; J. Cornuet; J. Marin & N.S. Pillai (2011). "Lack of confidence in approximate Bayesian computation model choice". Proceedings of the National Academy of Sciences. 108 (37): 15112–15117. Bibcode:2011PNAS..10815112R. doi: 10.1073/pnas.1102900108 . PMC   3174657 . PMID   21876135.
  13. Jeffreys, Harold (1998) [1961]. The Theory of Probability (3rd ed.). Oxford, England. p. 432. ISBN   9780191589676.{{cite book}}: CS1 maint: location missing publisher (link)
  14. Good, I.J. (1979). "Studies in the History of Probability and Statistics. XXXVII A. M. Turing's statistical work in World War II". Biometrika . 66 (2): 393–396. doi:10.1093/biomet/66.2.393. MR   0548210.
  15. Sharpening Ockham's Razor On a Bayesian Strop

Further reading