• Journal of the Korean Statistical Society
• /
• 제35권3호
• /
• pp.261-280
• /
• 2006

We have developed a new model selection criteria, the partial intrinsic Bayes factor, which is designed for cases when we select a model among a small number of candidate models. For example, we can choose only a few candidate models after exploring scatter plots. By simulation study, we have showed that PIBF performs better than AIC, BIC and GCV.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2011년도 추계학술대회
• /
• pp.208-209
• /
• 2011

본 논문에서는, 맵-리듀스 모델 기반에서 나이브 베이스 알고리즘으로 학습과 추론을 수행하는 방안에 대해 소개하고자 한다. 이를 위해 Apache Mahout를 이용하여 분산 나이브 베이스 (Distributed Naive Bayes) 학습 알고리즘을 University of California, Irvine (UCI)의 벤치마크 데이터 집합에 적용하였다. 실험 결과, Apache Mahout의 분산 나이브 베이스 학습 알고리즘은 일반적인 WEKA의 나이브 베이스 학습 알고리즘과 그 성능면에서 큰 차이가 없음을 알 수 있었다. 이러한 결과는, 향후 빅 데이터 환경에서 Apache Mahout와 같은 맵-리듀스 모델 기반 시스템이 기계 학습에 큰 기여를 할 수 있음을 나타내는 것이다.

• 응용통계연구
• /
• 제16권2호
• /
• pp.335-349
• /
• 2003

본 논문에서는 포아송분포 대 음이항분포, 그리고 정규분포, 이중지 수분포 대 코쉬분포에 대한 모형선택을 위하여 베이지안 방법을 사용한다. 각 모수에 대한 사전분포로는 무정보 부적절 사전분포의 가정 하에, 베이지안 모형선택을 위하여 O'Hagan (1995)의 부분적 베 이즈요인을 이용하였다. 실제자료와 모의 실험 자료의 분석을 통하여 부분적 베이즈요인의 유용성을 Berger와 Pericchi (1996, 1998)의 내재적 베이즈요인들과 함께 비교 검토해 본다.

• Communications for Statistical Applications and Methods
• /
• 제10권2호
• /
• pp.479-496
• /
• 2003

We investigate logit confidence intervals for the odds ratio based on the delta method. These intervals are constructed using pseudo-Bayes estimators. The Gart method and Agresti method smooth the observed counts toward the model of equiprobability and independence, respectively. We obtain better coverage probability by smoothing the observed counts toward the pseudo-Bayes estimators in 2$\times$2 table. We also improve legit confidence intervals in 2$\times$2$\times$K tables by generalizing these ideas. Utilizing pseudo-Bayes estimators, we obtain better coverage probability by smoothing the observed counts toward the conditional independence model, no three-factor interaction model and saturated model in 2$\times$2$\times$K tables.

• Communications for Statistical Applications and Methods
• /
• 제29권5호
• /
• pp.513-532
• /
• 2022

In this paper, the use of Bayes factors and the deviance information criterion for model selection are compared in a Bayesian accelerated life testing setup. In Bayesian accelerated life testing, the most used tool for model comparison is the deviance information criterion. An alternative and more formal approach is to use Bayes factors to compare models. However, Bayesian accelerated life testing models with more than one stressor often have mathematically intractable posterior distributions and Markov chain Monte Carlo methods are employed to obtain posterior samples to base inference on. The computation of the marginal likelihood is challenging when working with such complex models. In this paper, methods for approximating the marginal likelihood and the application thereof in the accelerated life testing paradigm are explored for dual-stress models. A simulation study is also included, where Bayes factors using the different approximation methods and the deviance information are compared.

• Communications for Statistical Applications and Methods
• /
• 제28권4호
• /
• pp.315-327
• /
• 2021

The Spatial Autoregressive (SAR) models have drawn considerable attention in recent econometrics literature because of their capability to model the spatial spill overs in a feasible way. While considering the Bayesian analysis of these models, one may face the problem of lack of robustness with respect to underlying prior assumptions. The generalized Bayes estimators provide a viable alternative to incorporate prior belief and are more robust with respect to underlying prior assumptions. The present paper considers the SAR model with a set of linear restrictions binding the regression coefficients and derives restricted generalized Bayes estimator for the coefficients vector. The minimaxity of the restricted generalized Bayes estimator has been established. Using a simulation study, it has been demonstrated that the estimator dominates the restricted least squares as well as restricted Stein rule estimators.

• Communications for Statistical Applications and Methods
• /
• 제7권1호
• /
• pp.141-150
• /
• 2000

In this paper we propose a method of computing Bayes factor to detect an outlier in a random effects model. When no information is available and hence improper noninformative priors should be used Bayes factor includes the unspecified constants and has complicated computational burden. To solve this problem we use the fractional Bayes factor (FBF) of O-Hagan(1995) and the generalized Savage0-Dickey density ratio of Verdinelli and Wasserman (1995) The proposed method is applied to outlier deterction problem We perform a simulation of the proposed approach with a simulated data set including an outlier and also analyze a real data set.

• 한국임상약학회지
• /
• 제23권2호
• /
• pp.137-141
• /
• 2013

Purpose: This study compared the performance of new NONMEM estimation methods using a population analysis dataset collected from a clinical study that consisted of 40 individuals and 567 observations after a single oral dose of glimepiride. Method: The NONMEM 7.2 estimation methods tested were first-order conditional estimation with interaction (FOCEI), importance sampling (IMP), importance sampling assisted by mode a posteriori (IMPMAP), iterative two stage (ITS), stochastic approximation expectation-maximization (SAEM), and Markov chain Monte Carlo Bayesian (BAYES) using a two-compartment open model. Results: The parameters estimated by IMP, IMPMAP, ITS, SAEM, and BAYES were similar to those estimated using FOCEI, and the objective function value (OFV) for diagnosing the model criteria was significantly decreased in FOCEI, IMPMAP, SAEM, and BAYES in comparison with IMP. Parameter precision in terms of the estimated standard error was estimated precisely with FOCEI, IMP, IMPMAP, and BAYES. The run time for the model analysis was shortest with BAYES. Conclusion: In conclusion, the new estimation methods in NONMEM 7.2 performed similarly in terms of parameter estimation, but the results in terms of parameter precision and model run times using BAYES were most suitable for analyzing this dataset.

• 응용통계연구
• /
• 제13권2호
• /
• pp.427-436
• /
• 2000

이 논문에서는 신뢰수명자료의 분석에 많이 사용되는 지수분포, 와이블분포, 로그정규분포에 대해, 현재의 자료가 어느 분포에 가장 적합한가를 알아보기 위한 베이자안 모형 선택방법을 제안한다. 일반적으로, 모수에 대한 사전분포가 부적절 분포인 경우, 베이즈 요인(Bayes factor)은 미지의 상수를 포함한다. 이러한 문제점을 해결하기 위하여 O’Hagan(1995)에 의해 제안된 fractional Bayes factor를 이용하여 자료를 가장 적합시키는 모형을 찾는다. 특히, 제2종 중도절단자료가 주어진 경우. 이 자료를 이용한 베이지안 모형선택에 대한 연구는 거의 이루어진 바가 없다. 실제 자료와 인위적인 자료를 이용하여 로그정규분포, 지수분포, 와이블모형중 어느 모형에 가장 잘 적합한지를 검정하는 예를 보인다.

• Journal of the Korean Data and Information Science Society
• /
• 제19권2호
• /
• pp.621-634
• /
• 2008

This paper addresses the problem of testing whether the means in several inverse Gaussian populations with heterogeneity are equal. The analysis of reciprocals for the equality of inverse Gaussian means needs the assumption of equal scale parameters. We propose Bayesian model selection procedures for testing equality of the inverse Gaussian means under the noninformative prior without the assumption of equal scale parameters. The noninformative prior is usually improper which yields a calibration problem that makes the Bayes factor to be defined up to a multiplicative constant. So we propose the objective Bayesian model selection procedures based on the fractional Bayes factor and the intrinsic Bayes factor under the reference prior. Simulation study and real data analysis are provided.