• Journal of the Korean Data and Information Science Society
• /
• v.11 no.2
• /
• pp.313-318
• /
• 2000

This article addresses aspects of combining information, with special attention to meta-analysis. In specific, we consider hierarchical Bayesian models for meta-analysis under priors which are scale mixtures of normal, and thus have tail heavier than that of the normal. Numerical methods of finding Bayes estimators under these heavy tailed prior are given, and are illustrated with an actual example.

• Management Science and Financial Engineering
• /
• v.12 no.2
• /
• pp.19-35
• /
• 2006

We use the hierarchical Bayesian approach to describe the transition probabilities of a binary nonhomogeneous Markov chain. The Markov chain is used for describing the transition behavior of emotionally disturbed children in a treatment program. The effects of covariates on transition probabilities are assessed using a logit link function. To describe the time evolution of transition probabilities, we consider two modeling strategies. The first strategy is based on the concept of exchangeabiligy, whereas the second one is based on a first order Markov property. The deviance information criterion (DIC) measure is used to compare models with two different time dependent structures. The inferences are made using the Markov chain Monte Carlo technique. The developed methodology is applied to some real data.

• Journal of KIISE:Software and Applications
• /
• v.31 no.1
• /
• pp.1-7
• /
• 2004

For the handwritten digit recognition, this paper Proposes a hierarchical Gator features extraction method and a Bayesian network for them. Proposed Gator features are able to represent hierarchically different level information and Bayesian network is constructed to represent hierarchically structured dependencies among these Gator features. In order to extract such features, we define Gabor filters level by level and choose optimal Gabor filters by using Fisher's Linear Discriminant measure. Hierarchical Gator features are extracted by optimal Gabor filters and represent more localized information in the lower level. Proposed methods were successfully applied to handwritten digit recognition with well-known naive Bayesian classifier, k-nearest neighbor classifier. and backpropagation neural network and showed good performance.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2007.05a
• /
• pp.1752-1756
• /
• 2007

정교한 강우-유출 모의를 위해서는 적절한 매개변수의 추정이 필수적이며, 매개변수 추정 방법은 시행착오(trial and error)에 의한 수동보정법과 최적화방법을 사용한 자동보정법으로 구분할 수 있다. 모형의 매개변수의 수가 많은 경우 수동보정법에 의한 매개변수 추정은 매우 어렵다. 자동 보정법에 사용되는 최적화방법은 Rosenbrock 알고리즘, patten search, 컴플렉스(complex) 방법, Powell 방법 등과 같은 지역최적화 방법과 전역최적화 방법으로 나눌 수 있다. 그러나 기존 방법론들은 매개변수의 최적화를 추적하기 위한 알고리즘이 대부분이며 이들 매개변수에 관련된 불확실성을 평가하는데는 미흡한 단접이 있다. 이러한 점에서 본 연구에서는 강우-유출모형의 매개변수 추정에 있어서 불확실성을 평가할 수 있는 새로운 방법론을 검토하고자 한다. 매개변수와 관련된 불확실성을 평가하기 위한 방법은 여러 가지가 있으나 통계적으로 매우 우수한 능력을 보이는 Hierarchical Bayesian 알고리즘을 Probability-Distributed 강우-유출 모형에 적용하였다. 본 방법론은 최적화와 동시에 각 매개변수에 관련된 사후분포(posterior distribution)의 추정이 가능하므로 모형이 갖는 불확실성을 효과적으로 평가할 수 있다. 따라서, 수자원 관리에 있어서 불확실성을 고려할 수 있으므로 보다 수리수문학적 위험도를 저감할 수 있을 것으로 판단된다.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2016.05a
• /
• pp.206-206
• /
• 2016

최근 기상변동성 증가 및 기후변화로 인해 기존 한반도의 기상패턴과 다른 이상강우 현상이 증가하고 있다. 이러한 변동성 증가는 수자원 계획을 수립하는데 있어 불확실성을 가중시키기고 있다. 이러한 점에서 수문 시계열의 변화양상을 효과적으로 인지할 수 있으며, 유역단위에서 일관된 변화를 평가할 수 있는 변동성 분석 개발이 필요하다. 이에 본 연구에서는 기존 변동성 분석방법에 계층적 베이지안(Hierarchical Bayesian) 기법을 연계하여 유역단위에서 변동점 해석을 위한 모형을 개발하였다. 한강유역의 30년 이상의 강우 자료를 활용하여 연강우량 자료를 구축하였으며, 본 연구를 통해 개발된 모형의 적합성을 평가하였다. 분석결과, 약 2000년대를 기준으로 강우의 변화 양상을 확인할 수 있었으며, 과거에 비해 강우의 증가 특성을 효과적으로 평가할 수 있었다. 이와 같은 수문기상자료에 대한 변동성 분석은 미래에 발생 가능한 홍수나 가뭄과 같은 사상을 모의함에 있어 효율적으로 활용될 수 있을 것으로 판단된다.

• Journal of the Korean Statistical Society
• /
• v.28 no.2
• /
• pp.167-182
• /
• 1999

This article is concerned with the selection of subsets of predictor variables to be included in building the binary response probit regression model. It is based on a Bayesian approach, intended to propose and develop a procedure that uses probabilistic considerations for selecting promising subsets. This procedure reformulates the probit regression setup in a hierarchical normal mixture model by introducing a set of hyperparameters that will be used to identify subset choices. The appropriate posterior probability of each subset of predictor variables is obtained through the Gibbs sampler, which samples indirectly from the multinomial posterior distribution on the set of possible subset choices. Thus, in this procedure, the most promising subset of predictors can be identified as the one with highest posterior probability. To highlight the merit of this procedure a couple of illustrative numerical examples are given.

• 한국HCI학회:학술대회논문집
• /
• 2009.02a
• /
• pp.6-10
• /
• 2009

In these days, recommendation service in mobile environments is in the limelight due to the spread of mobile devices and an increase of information owing to advancement of computer network. The restaurant recommendation system reflecting user preference was proposed. This system uses Bayesian network to model user preference and analytical hierarchical process to recommend restaurants, but static inference model for user preference used in the system has some limitations that cannot manage changing user preference and enormous user survey must be preceded. This paper proposes a learning method for Bayesian network based on user requests. The proposed method is implemented on mobile devices and desktop, and we show the possibility of the proposed method through experiments.

• Journal of the Korean Data and Information Science Society
• /
• v.17 no.3
• /
• pp.1009-1019
• /
• 2006

In this paper, we consider the simultaneous estimation problem for the normal means. We set up the model structure using the several different distributions of the errors for observing their effects of model perturbation for the error terms in obtaining the empirical Bayes and hierarchical Bayes estimators. We compare the performance of those estimators under model perturbation based on a simulation study.

• Proceedings of the Korean Operations and Management Science Society Conference
• /
• 2005.05a
• /
• pp.798-805
• /
• 2005

Estimating the effects of price increase on a company's sales is important task faced by managers. If consumer has prior information on price increase or expect it, there would be stockpiling and subsequent drops in sales. In addition, consumer can suppress demand in the short run. Above factors make the sales dynamic and unstable. We develop a time series model to evaluate the sales patterns with stockpiling and short term suppression of demand and also propose a forecasting procedure. For estimation, we use panel data and extend the model to Bayesian hierarchical structure. By borrowing strength across cross-sectional units, this estimation scheme gives more robust and reasonable result than one from the individual estimation. Furthermore, the proposed scheme yields improved predictive power in the forecasting of hold-out sample periods.

• Journal of the Korean Statistical Society
• /
• v.29 no.4
• /
• pp.407-422
• /
• 2000

This article is concerned with the selecting predictor variables to be included in building a class of binary response t-link regression models where both probit and logistic regression models can e approximately taken as members of the class. It is based on a modification of the stochastic search variable selection method(SSVS), intended to propose and develop a Bayesian procedure that used probabilistic considerations for selecting promising subsets of predictor variables. The procedure reformulates the binary response t-link regression setup in a hierarchical truncated normal mixture model by introducing a set of hyperparameters that will be used to identify subset choices. In this setup, the most promising subset of predictors can be identified as that with highest posterior probability in the marginal posterior distribution of the hyperparameters. To highlight the merit of the procedure, an illustrative numerical example is given.