• Journal of the Korea Safety Management & Science
• /
• v.11 no.4
• /
• pp.119-125
• /
• 2009

We predict the earthquake rate in Korea following Bayesian approach. We make a model that can utilize the data to predict other levels of earthquake. An event tree model which is a frequently used graphical tool in describing accident initiation and escalation to more severe accident is transformed into an influence diagram model. Prior distributions for earthquake occurrence rate and probabilities to escalating to more severe earthquakes are assumed and likelihood of number of earthquake in a given period of time is assessed. And then posterior distributions are obtained based on observed data. We find that the minor level of earthquake is increasing while major level of earthquake is less likely.

• Communications for Statistical Applications and Methods
• /
• v.9 no.3
• /
• pp.835-844
• /
• 2002

For the one-way random model when the ratio of the variance components is of interest, Bayesian analysis is often appropriate. In this paper, we develop the noninformative priors for the ratio of the variance components under the balanced one-way random effect model. We reveal that the second order matching prior matches alternative coverage probabilities up to the second order (Mukerjee and Reid, 1999) and is a HPD(Highest Posterior Density) matching prior. It turns out that among all of the reference priors, the only one reference prior (one-at-a-time reference prior) satisfies a second order matching criterion. Finally we show that one-at-a-time reference prior produces confidence sets with expected length shorter than the other reference priors and Cox and Reid (1987) adjustment.

• Communications for Statistical Applications and Methods
• /
• v.24 no.5
• /
• pp.457-480
• /
• 2017

We develop a random partition procedure based on a Dirichlet process prior with Laplace distribution. Gibbs sampling of a Laplace mixture of linear mixed regressions with a Dirichlet process is implemented as a random partition model when the number of clusters is unknown. Our approach provides simultaneous partitioning and parameter estimation with the computation of classification probabilities, unlike its counterparts. A full Gibbs-sampling algorithm is developed for an efficient Markov chain Monte Carlo posterior computation. The proposed method is illustrated with simulated data and one real data of the energy efficiency of Tsanas and Xifara (Energy and Buildings, 49, 560-567, 2012).

• Proceedings of the Korea Water Resources Association Conference
• /
• 2009.05a
• /
• pp.1147-1151
• /
• 2009

In this study, we address the problem of producing probability forecasts of summer seasonal rainfall, on the basis of Hindcast experiments from a ensemble of GCMs(cwb, gcps, gdaps, metri, msc_gem, msc_gm2, msc_gm3, msc_sef and ncep). An advanced Hierarchical Bayesian weighting scheme is developed and used to combine nine GCMs seasonal hindcast ensembles. Hindcast period is 23 years from 1981 to 2003. The simplest approach for combining GCM forecasts is to weight each model equally, and this approach is referred to as pooled ensemble. This study proposes a more complex approach which weights the models spatially and seasonally based on past model performance for rainfall. The Bayesian approach to multi-model combination of GCMs determines the relative weights of each GCM with climatology as the prior. The weights are chosen to maximize the likelihood score of the posterior probabilities. The individual GCM ensembles, simple poolings of three and six models, and the optimally combined multimodel ensemble are compared.

• Proceedings of the KSRS Conference
• /
• 2005.10a
• /
• pp.430-433
• /
• 2005

In this study, weight of evidence(WOE) technique based on the bayesian method was applied to estimate the groundwater yield characteristics in the Pocheon area in Kyungki-do. The ground water preservation depends on many hydrogeologic factors that include hydrologic data, landuse data, topographic data, geological map and other natural materials, even with man-made things. All these data can be digitally collected and managed by GIS database. In the applied technique of WOE, The prior probabilities were estimated as the factors that affect the yield on lineament, geology, drainage pattern or river system density, landuse and soil. We calculated the value of the Weight W+, W- of each factor and estimated the contrast value of it. Results by the ground water yield characteristic calculations were presented in the form of posterior probability map to the consideration of in-situ samples. It is concluded that this technique is regarded as one of the effective technique for the feasibility mapping related to detection of groundwater bearing zones and its spatial pattern.

• Communications for Statistical Applications and Methods
• /
• v.18 no.4
• /
• pp.467-475
• /
• 2011

This paper develops the noninformative priors for the stress-strength reliability from one parameter generalized exponential distributions. When this reliability is a parameter of interest, we develop the first, second order matching priors, reference priors in its order of importance in parameters and Jeffreys' prior. We reveal that these probability matching priors are not the alternative coverage probability matching prior or a highest posterior density matching prior, a cumulative distribution function matching prior. In addition, we reveal that the one-at-a-time reference prior and Jeffreys' prior are actually a second order matching prior. We show that the proposed reference prior matches the target coverage probabilities in a frequentist sense through a simulation study and a provided example.

• 한국데이터정보과학회:학술대회논문집
• /
• 2005.10a
• /
• pp.71-84
• /
• 2005

The Fieller-Creasy problem involves statistical inference about the ratio of two independent normal means. It is difficult problem from either a frequentist or a likelihood perspective. As an alternatives, a Bayesian analysis with noninformative priors may provide a solution to this problem. In this paper, we extend the results of Yin and Ghosh (2001) to unbalanced sample case. We find various noninformative priors such as first and second order matching priors, reference and Jeffreys' priors. The posterior propriety under the proposed noninformative priors will be given. Using real data, we provide illustrative examples. Through simulation study, we compute the frequentist coverage probabilities for probability matching and reference priors. Some simulation results will be given.

• Journal of the Korean Data and Information Science Society
• /
• v.18 no.2
• /
• pp.553-560
• /
• 2007

In this paper, we consider two components system which lifetimes have Freund's bivariate exponential model with equal failure rates. We propose Bayesian multiple comparisons procedure for the failure rates of I Freund's bivariate exponential populations based on Dirichlet process priors(DPP). The family of DPP is applied in the form of baseline prior and likelihood combination to provide the comparisons. Computation of the posterior probabilities of all possible hypotheses are carried out through Markov Chain Monte Carlo(MCMC) method, namely, Gibbs sampling, due to the intractability of analytic evaluation. The whole process of multiple comparisons problem for the failure rates of bivariate exponential populations is illustrated through a numerical example.

• Journal of the Korean Data and Information Science Society
• /
• v.21 no.5
• /
• pp.935-944
• /
• 2010

In this paper, we develop the noninformative priors for the reliability in a stress-strength model where a strength X and a stress Y have independent exponential distributions with different scale parameters and a common location parameter. We derive the reference priors and prove the propriety of joint posterior distribution under the general prior including the reference priors. Through the simulation study, we show that the proposed reference priors match the target coverage probabilities in a frequentist sense.

• MALSORI
• /
• no.53
• /
• pp.103-118
• /
• 2005

The performance of speech recognition system is degraded by mismatch between training and test environments. Recently, Stereo-based Piecewise LInear Compensation for Environments (SPLICE) was introduced to overcome environmental mismatch using stereo data. In this paper, we propose several methods to improve the conventional SPLICE and evaluate them in the Aurora2 task. We generalize SPLICE to compensate for covariance matrix as well as mean vector in the feature space, and thereby yielding the error rate reduction of 48.93%. We also employ the weighted sum of correction vectors using posterior probabilities of all Gaussians, and the error rate reduction of 48.62% is achieved. With the combination of the above two methods, the error rate is reduced by 49.61% from the Aurora2 baseline system.