• Communications for Statistical Applications and Methods
• /
• 제24권6호
• /
• pp.543-559
• /
• 2017

Tree-based regression and classification ensembles form a standard part of the data-science toolkit. Many commonly used methods take an algorithmic view, proposing greedy methods for constructing decision trees; examples include the classification and regression trees algorithm, boosted decision trees, and random forests. Recent history has seen a surge of interest in Bayesian techniques for constructing decision tree ensembles, with these methods frequently outperforming their algorithmic counterparts. The goal of this article is to survey the landscape surrounding Bayesian decision tree methods, and to discuss recent modeling and computational developments. We provide connections between Bayesian tree-based methods and existing machine learning techniques, and outline several recent theoretical developments establishing frequentist consistency and rates of convergence for the posterior distribution. The methodology we present is applicable for a wide variety of statistical tasks including regression, classification, modeling of count data, and many others. We illustrate the methodology on both simulated and real datasets.

• Journal of the Korean Data and Information Science Society
• /
• 제26권1호
• /
• pp.261-269
• /
• 2015

Nowadays the application of change point analysis has been indispensable in a wide range of areas such as quality control, finance, environmetrics, medicine, geographics, and engineering. Identification of times where process changes would help minimize the consequences that might happen afterwards. The main objective of this paper is to compare the change-point detection capabilities of Bayesian estimate and maximum likelihood estimate. We applied Bayesian and maximum likelihood techniques to formulate change points having a step change and multiple number of change points in a Poisson rate. After a signal from c-chart and Poisson cumulative sum control charts have been detected, Monte Carlo simulation has been applied to investigate the performance of Bayesian and maximum likelihood estimation. Change point detection capacities of Bayesian and maximum likelihood estimation techniques have been investigated through simulation. It has been found that the Bayesian estimates outperforms standard control charts well specially when there exists a small to medium size of step change. Moreover, it performs convincingly well in comparison with the maximum like-lihood estimator and remains good choice specially in confidence interval statistical inference.

• 한국지구물리탐사학회:학술대회논문집
• /
• 한국지구물리탐사학회 2003년도 Proceedings of the international symposium on the fusion technology
• /
• pp.340-343
• /
• 2003

This study presents a practical procedure for the Bayesian inversion of geophysical data by Markov chain Monte Carlo (MCMC) sampling and geostatistics. We have applied geostatistical techniques for the acquisition of prior model information, and then the MCMC method was adopted to infer the characteristics of the marginal distributions of model parameters. For the Bayesian inversion of dipole-dipole array resistivity data, we have used the indicator kriging and simulation techniques to generate cumulative density functions from Schlumberger array resistivity data and well logging data, and obtained prior information by cokriging and simulations from covariogram models. The indicator approach makes it possible to incorporate non-parametric information into the probabilistic density function. We have also adopted the MCMC approach, based on Gibbs sampling, to examine the characteristics of a posteriori probability density function and the marginal distribution of each parameter. This approach provides an effective way to treat Bayesian inversion of geophysical data and reduce the non-uniqueness by incorporating various prior information.

• 한국신뢰성학회지:신뢰성응용연구
• /
• 제14권3호
• /
• pp.162-168
• /
• 2014

This paper discusses the application of Bayesian techniques with test data on similar products for performing the Economical Reliability Test of new one-shot device. Using the test data on similar products, reliability test required lower sample size currently being spent in order to demonstrate a target reliability with a specified confidence level. Furthermore, lower sample size reduces cost, time and various resources on reliability test. In this paper, we use similarity as calculating weight of similar products and analyze similarity between new and similar product for comparison of the essential function.

• 디지털융복합연구
• /
• 제13권8호
• /
• pp.295-300
• /
• 2015

사전 확률분포를 모델링하는 HMM을 사용하는 어휘 인식에서 인식 어휘의 모델들의 대한 인식 확률이 이산적인 분포를 나타내며 인식을 위한 계산량이 적은 장점이 있지만 인식률을 계산했을 때 상대적으로 낮은 단점이 있다. 이를 개선하기 위하여 베이시안 기법 어휘 인식 모델을 융합한 잡음 제거 인식률 향상을 제안한다. 본 논문은 베이시안 기법 어휘 인식을 위한 모델 구성을 베이시안 기법의 최적화한 인식 모델을 구성하였다. HMM을 기반으로 한 사전 확률 방법과 베이시안 기법인 사후확률을 융합하여 잡음을 제거하고 인식률을 향상시켰다. 본 논문에서 제안한 방법을 적용한 결과 어휘 인식률에서 98.1%의 인식률을 나타내었다.

• 품질경영학회지
• /
• 제24권1호
• /
• pp.61-72
• /
• 1996

It is very important to determine the likelihood ratios and the prior odds for designing a Bayesian processor in Built-In-Test system. Using traditional statistics, it is not difficult to determine the initial prior odds from the field data. For a newly designed system, development testing data or laboratory testing data could be used to replace field data. The likelihood ratios which playa key role in the Bayesian processor must be carefully determined, based on laboratory testing and statistical techniques. In this paper, expressing and determining the likelihood ratios by Geometric areas, Test, and Analytical method will be presented.

• Journal of the Korean Data and Information Science Society
• /
• 제15권1호
• /
• pp.227-237
• /
• 2004

In this paper we presented a Bayesian inference approach for estimating the location and scale parameters of the lognormal distribution using iterative Gibbs sampling algorithm. We also presented estimation of location parameter by two non iterative methods, importance sampling and weighted bootstrap assuming scale parameter as known. The estimates by non iterative techniques do not depend on the specification of hyper parameters which is optimal from the Bayesian point of view. The estimates obtained by more sophisticated Gibbs sampler vary slightly with the choices of hyper parameters. The objective of this paper is to illustrate these tools in a simpler setup which may be essential in more complicated situations.

• Journal of the Korean Statistical Society
• /
• 제19권1호
• /
• pp.71-79
• /
• 1990

In the linear model which consider both the multivariate parallel-line bioassay and the multivariate linear calibration, this paper presents a Bayesian procedure which is an extension of Hunter and Lamboy (1981) and has several advantages compared with the non Bayesian techniques. Based on the methods of this article we discuss the effect of multivariate calibration and give a numerical example.

• 품질경영학회지
• /
• 제42권4호
• /
• pp.747-756
• /
• 2014

Purpose: The purpose of this study is to apply the Bayesian estimation methodology for producing 'Korean Standard -Quality Excellence Index' model and prove the effectiveness of the new approach based on survey data by comparing the current index with the new index produced by Bayesian estimation method. Methods: The 'Korean Standard -Quality Excellence Index' was produced through the collected survey data by Bayesian estimation method and comparing the deviation with two results for confirming the effectiveness of suggested application. Results: The statistical analysis result shows that suggested estimator, that is, empirical Bayes estimator improves the effectiveness of the index with regard to reduce the error under specific loss function, which is suggested for checking the goodness of fit. Conclusion: Considering the Bayesian techniques such as empirical Bayes estimator for producing the quality excellence index reduces the error for estimating the parameter of interest and furthermore various Bayesian perspective approaches seems to be meaningful for producing the corresponding index.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2016년도 학술발표회
• /
• pp.528-528
• /
• 2016

신뢰성 있는 수문순환모의를 위해서 다양한 수문모형이 사용되고 있다. 그 중 대표적인 수문모형인 강우-유출 모형은 유역에 발생한 강우에 반응하는 유출특성을 평가하는데 이용된다. 강우-유출 과정은 강우량, 유출량, 도달시간 및 토양수분 등과 연관된 매개변수들의 최적화 과정을 통해서 추정된다. 하지만 동일한 강우사상이라도 다양한 매개변수들로 인하여 상당히 다른 유출패턴을 나타내기 때문에 수문순환 과정을 정확히 모의하기 위해서 강우-유출 분석시 불확실성 분석이 필수적으로 요구된다. 불확실성 분석은 통계학에서도 쉽지 않은 연구내용으로서 가장 진보된 불확실성 분석기법인 Bayesian 기법은 매개변수의 추정과 불확실성 분석을 동시에 수행할 수 있는 방법으로 매개변수들은 사후분포(posterior distribution)로 귀결되며 최종적으로 확률분포형의 형태를 가진다. 본 연구에서는 국내외적으로 널리 사용되는 단기유출 모형 HEC-1 모형에 Bayesian 기법을 연계하여 대상유역의 도달시간, 저류상수 및 CN No. 최적화 및 불확실성 평가를 수행하였다. 연구결과 Bayesian 기법을 통한 매개변수 최적화 결과는 안정적인 수렴결과를 확인하였으며, 확률강우량을 입력자료로 사용하여 산정된 빈도별 홍수수분곡선의 불확실성 분석을 통하여 향후 수공구조물의 위험도 분석 및 수자원계획 수립시 유용한 자료로 사용될 것으로 판단된다.