翻訳と辞書 Words near each other ・ Devi Sharma (politician) ・ Devi Shetty ・ Devi Singh ・ Devi Singh Bhati ・ Devi Sri Prasad ・ Devi sukta ・ Devi Upanishad ・ Devi, Khyber Pakhtunkhwa ・ Devi, Punjab ・ Devi-Bhagavata Purana ・ Devia ・ Devian ・ Deviance ・ Deviance (sociology) ・ Deviance (statistics) ・ Deviance information criterion ・ Deviance regulation theory ・ Deviancy amplification spiral ・ Deviant (comics) ・ Deviant (disambiguation) ・ Deviant (Pitchshifter album) ・ Deviant (Regurgitate album) ・ Deviant behavior ・ Deviant Behavior (journal) ・ Deviant logic ・ Deviant Records ・ Deviant sexual intercourse ・ DeviantArt ・ Devianz ・ Deviat Dictionary Lists mini英和辞書 mini和英辞書 Webster 1913 Latin-English FOLDOC Wikipedia English ウィキペディア

翻訳と辞書　辞書検索 [ 開発暫定版 ] スポンサード リンク Deviance information criterion ： ウィキペディア英語版 Deviance information criterion The deviance information criterion (DIC) is a hierarchical modeling generalization of the AIC (Akaike information criterion) and BIC (Bayesian information criterion, also known as the Schwarz criterion). It is particularly useful in Bayesian model selection problems where the posterior distributions of the models have been obtained by Markov chain Monte Carlo (MCMC) simulation. Like AIC and BIC it is an asymptotic approximation as the sample size becomes large. It is only valid when the posterior distribution is approximately multivariate normal. Define the deviance as $D(\backslash theta)=-2\; \backslash log(p(y|\backslash theta))+C\backslash ,$, where $y\backslash ,$ are the data, $\backslash theta\backslash ,$ are the unknown parameters of the model and $p(y|\backslash theta)\backslash ,$ is the likelihood function. $C\backslash ,$ is a constant that cancels out in all calculations that compare different models, and which therefore does not need to be known. The expectation $\backslash bar=\backslash mathbf^\backslash theta()$ is a measure of how well the model fits the data; the larger this is, the worse the fit. There are two calculations in common usage for the effective number of parameters of the model. The first, as described in is $p\_D=\backslash bar-D(\backslash bar)$, where $\backslash bar$ is the expectation of $\backslash theta\backslash ,$. The second, as described in is $p\_D\; =\; p\_V\; =\; \backslash frac\backslash widehat\; =\; p\_D+\backslash bar,$ or equivalently as :$\backslash mathit\; =\; D(\backslash bar)+2\; p\_D.$ From this latter form, the connection with Akaike's information criterion is evident. The idea is that models with smaller DIC should be preferred to models with larger DIC. Models are penalized both by the value of $\backslash bar$, which favors a good fit, but also (in common with AIC and BIC) by the effective number of parameters $p\_D\backslash ,$. Since $\backslash bar\; D$ will decrease as the number of parameters in a model increases, the $p\_D\backslash ,$ term compensates for this effect by favoring models with a smaller number of parameters. The advantage of DIC over other criteria in the case of Bayesian model selection is that the DIC is easily calculated from the samples generated by a Markov chain Monte Carlo simulation. AIC and BIC require calculating the likelihood at its maximum over $\backslash theta\backslash ,$, which is not readily available from the MCMC simulation. But to calculate DIC, simply compute $\backslash bar$ as the average of $D(\backslash theta)\backslash ,$ over the samples of $\backslash theta\backslash ,$, and $D(\backslash bar)$ as the value of $D\backslash ,$ evaluated at the average of the samples of $\backslash theta\backslash ,$. Then the DIC follows directly from these approximations. Claeskens and Hjort (2008, Ch. 3.5) show that the DIC is large-sample equivalent to the natural model-robust version of the AIC. In the derivation of DIC, it is assumed that the specified parametric family of probability distributions that generate future observations encompasses the true model. This assumption does not always hold, and it is desirable to consider model assessment procedures in that scenario. Also, the observed data are used both to construct the posterior distribution and to evaluate the estimated models. Therefore, DIC tends to select over-fitted models. Recently, these issues are resolved by , Bayesian predictive information criterion, BPIC. Ando (2010, Ch. 8) provided a discussion of various Bayesian model selection criteria. To avoid the over-fitting problems of DIC, developed Bayesian model selection criteria from a predictive view point. The criterion is calculated as :$\backslash mathit\; =\backslash bar+2p\_D=-2\backslash mathbf^\backslash theta(\backslash log(p(y|\backslash theta)))+2p\_D.$ The first term is a measure of how well the model fits the data, while the second term is a penalty on the model complexity. Note, that the p in this expression is the predictive distribution rather than the likelihood above. ==See also== * Akaike information criterion (AIC) * Bayesian information criterion (BIC) * Bayesian predictive information criterion (BPIC) * Focused information criterion (FIC) * Hannan-Quinn information criterion * Kullback–Leibler divergence * Jensen–Shannon divergence 抄文引用元・出典: フリー百科事典『 ウィキペディア（Wikipedia）』 ■ウィキペディアで「Deviance information criterion」の詳細全文を読む スポンサード リンク 翻訳と辞書 : 翻訳のためのインターネットリソース Copyright(C) kotoba.ne.jp 1997-2016. All Rights Reserved.