• 품질경영학회지
• /
• 제34권1호
• /
• pp.48-53
• /
• 2006

Applying Bayesian optimal design principles is not easy when a prior distribution is not certain. We present a optimal design criterion which possibly yield a reasonably good design and also robust with respect to misspecification of the prior distributions. The criterion is applied to the problem of estimating the turning point of a quadratic regression. Exact mathematical results are presented under certain conditions on prior distributions. Computational results are given for some cases not satisfying our conditions.

• Communications for Statistical Applications and Methods
• /
• 제15권1호
• /
• pp.95-104
• /
• 2008

This paper addresses issues of robustness in Bayesian optimal design. We may have difficulty applying Bayesian optimal design principles because of the uncertainty of prior distribution. When there are several plausible prior distributions and the efficiency of a design depends on the unknown prior distribution, robustness with respect to misspecification of prior distribution is required. We suggest a new optimal design criterion which has relatively high efficiencies across the class of plausible prior distributions. The criterion is applied to the problem of estimating the turning point of a quadratic regression, and both analytic and numerical results are shown to demonstrate its robustness.

• Journal of the Korean Statistical Society
• /
• 제30권1호
• /
• pp.163-178
• /
• 2001

We consider the experimental design problem of selecting values of design variables x for observation of a response y that depends on x and on model parameters $\theta$. The form of the dependence may be quite general, including all linear and nonlinear modeling situations. The goal of the design selection is to efficiently estimate functions of $\theta$. Three new criteria for selecting design points x are presented. The criteria generalized the usual Bayesian optimal design criteria to situations n which the prior distribution for $\theta$ amy be uncertain. We assume that there are several possible prior distributions,. The new criteria are applied to the nonlinear problem of designing to estimate the turning point of a quadratic equation. We give both analytic and computational results illustrating the robustness of the optimal designs based on the new criteria.

• 응용통계연구
• /
• 제10권1호
• /
• pp.69-84
• /
• 1997

선형 캘리브레이션 실험계획 문제에 대하여, 베이지안 의사결정론을 이용하여 평균제곱오차손실을 최소화한 Kim(1988, 1993)의 실험계획과 관련 문헌의 결과인 몇 가지 최적계획을 비교한다. 비교대상 실험계획으로서 고전적 추정량의 점근분산을 최소화하는 Buonaccorsi(1986)의 최적계획, 회귀분석 모형에서 $ M(x) = \sum x_i x_i '$의 함수를 최대화 또는 최소화하는 D-optimal 또는 A-optimal 계획, Hunter and Lamboy(1981)가 베이지안 추정량의 특성을 설명하기 위하여 그 논문에서 예로 들었던 실험계획을 고려한다. 서로 다른 기준에 의한 최적계획을 비교하기 위해서 우선 기대사후분산을 계산하여 비교하고 몇가지 사전분포에 대하여 몬테칼로 시뮬레이션을 통한 평균분산과 HPD 구간의 크기를 비교한다.

• 응용통계연구
• /
• 제27권6호
• /
• pp.1029-1038
• /
• 2014

본 연구에서는 베이지안 실험계획법에 대하여 논의하고 간단한 모의실험을 통하여 최적화된 베이지안 실험계획법이 어떠한 특징을 가지고 있는지 설명하였다. 실험을 설계하는 경우 연구자는 관심있는 주제가 모수추정인지 아니면 예측인지를 결정하고 사전확률과 우도함수를 기반으로 이에 맞는 사후확률을 찾아 효용함수와 결합하여 최적의 실험설계를 찾는 것이 베이지안 실험계획법의 기본 원리이다. 만일 사전적 정보가 존재하지 않는다면 무정보적 부적합 사전확률을 이용하여 실험을 설계할 수 있으며, 이는 비 베이지안적 접근방법과 일치하게 된다. 만일 모수나 예측값에 대한 사전적 정보가 존재하는 경우에는 베이지안 실험계획법이 유일한 해결 방법이다. 하지만 모형의 복잡도가 증가하게 되면, 최적해를 찾는 과정이 매우 복잡해져서 극복해야 하는 많은 문제점들이 존재하므로 향후 많은 연구가 필요한 분야이다.

• 대한수학회지
• /
• 제31권4호
• /
• pp.709-715
• /
• 1994

The aim of of the study is a powerful test for the discrimination and therefore an optimal desin for that purpose. This problem is studied by Chernoff ([5]) and used in Chernoff ([6]) for accelerated life tests using the exponential distribution for life times. The approach used here is similar to that suggested by Lauter ([10]) and used in Chaloner ([3]) and Chaloner and Larntz ([4]) where it is motivated using Bayesian arguments. The approach taken in this paper the loss function $L(\cdot)$ evaluating a test procedure and a design d.

• Journal of the Korean Statistical Society
• /
• 제22권2호
• /
• pp.219-234
• /
• 1993

A measuring instrument must be calibrated for accurate inferences of an unknown quantity. Bayesian calibration designs with respect to squared error loss based on a linear model are discussed in Kim and Barlow (1992). In this paper, we consider approximations of the optimal calibration designs using the idea of Gaussian inflence diagrams. The approximation is evaluated by means of numerical calculations, where it is compared with the exact values from the numerical integration.

• 한국경영과학회지
• /
• 제20권3호
• /
• pp.105-122
• /
• 1995

Based on linear models, the inference about the true measurement x$_{f}$ and the optimal designs x (nx1) for the calibration experiments are considered via Baysian statistical decision analysis. The posterior distribution of x$_{f}$ given the observation y$_{f}$ (qxl) and the calibration experiment is obtained with normal priors for x$_{f}$ and for themodel parameters (.alpha., .betha.). This posterior distribution is not in the form of any known distributions, which leads to the use of a numerical integration or an approximation for the calculation of the overall expected loss. The general structure of the expected loss function is characterized in the form of a conjecture. A near-optimal design is obtained through the approximation nof the conditional covariance matrix of the joint distribution of (x$_{f}$ , y$_{f}$ $^{T}$ )$^{T}$ . Numerical results for the univariate case are given to demonstrate the conjecture and to evaluate the approximation.n.

• Journal of Electrical Engineering and Technology
• /
• 제2권4호
• /
• pp.543-548
• /
• 2007

We present a dynamic Bayesian network (DBN) model of a generalized class of nonstationary birth-death processes. The model includes birth and death rate parameters that are randomly selected from a known discrete set of values. We present an on-line algorithm to obtain optimal estimates of the parameters. We provide a simulation of real-time characterization of load traffic estimation using our DBN approach.

• Communications for Statistical Applications and Methods
• /
• 제6권1호
• /
• pp.207-220
• /
• 1999

In this paper, some of the issue about a group sequential method are considered in the Bayesian context. The continuous time optimal stopping boundary can be used to approximate the optimal stopping boundary for group sequential designs. The exact stopping boundary for group sequential design is obtained by using the backward induction method and is compared with the continuous optimal stopping boundary and the corrected continuous stopping boundary.