• Communications for Statistical Applications and Methods
• /
• 제12권2호
• /
• pp.483-496
• /
• 2005

We consider the power law process which is assumed to have multiple changepoints. We propose a binary segmentation procedure for locating all existing changepoints. We select one model between the no-changepoints model and the single changepoint model by the Bayes factor. We repeat this procedure until no more changepoints are found. Then we carry out a multiple test based on the Bayes factor through the intrinsic priors of Berger and Pericchi (1996) to investigate the system behaviour of failure times. We demonstrate our procedure with a real dataset and some simulated datasets.

• Journal of the Korean Statistical Society
• /
• 제32권3호
• /
• pp.271-288
• /
• 2003

In this paper, when the variance of the normal distribution is related to the mean, we develop noninformative priors such as matching priors and reference priors. We prove that the second order matching prior matches alternative coverage probabilities up to the same order and also it is a HPD matching prior. It turns out that one-at-a-time reference prior satisfies a second order matching criterion. Then using these noninformative priors, we develop a Bayesian test procedure for the equality of the means and variances of two independent normal distributions using fractional Bayes factor. Some simulation study is performed, and a real data example is also provided.

• 한국품질경영학회:학술대회논문집
• /
• 한국품질경영학회 2006년도 춘계학술대회
• /
• pp.319-324
• /
• 2006

In this paper, we propose Bayesian procedure for the multiple change points analysis in a sequence of fractions nonconforming. We first compute the Bayes factor for detecting the existence of no change, a single change or multiple changes. The Gibbs sampler with the Metropolis-Hastings subchain is run to estimate parameters of the change point model, once the number of change points is identified. Finally, we apply the results developed in this paper to both a real and simulated data.

• International Journal of Reliability and Applications
• /
• 제14권2호
• /
• pp.55-70
• /
• 2013

This paper deals with the Bayesian analysis of the failure data of a repairable mechanical system subject to minimal repairs and periodic overhauls. The effect of overhauls on the reliability of the system is modeled by a proportional age reduction model and the failure process between two successive overhauls is assumed to be 2-parameter Engelhardt-Bain process (2-EBP). Power Law Process (PLP) model has a disadvantage which 2-EBP can overcome. On the basis of the observed data and of a number of suitable prior densities, point and interval estimation of model parameters, as well as quantities of relevant interest are found. Also hypothesis tests on the effectiveness of performed overhauls have been developed using Bayes factor. Sensitivity analysis of improvement parameter is carried out. Finally, a numerical application is used to illustrate the proposed method.

• 한국정보기술응용학회:학술대회논문집
• /
• 한국정보기술응용학회 2001년도 추계공동학술대회 논문자료집 정보화 젼략 패러다임의 변화에 대한 보기술의 대응
• /
• pp.155-155
• /
• 2001

• 품질경영학회지
• /
• 제24권4호
• /
• pp.190-206
• /
• 1996

In case measurements are made on units of production in time order, it is reasonable to expect that the measurement errors will sometimes be first order autocorrelated, and a technique to test such autocorrelation is required to give good control of the productive process. Tool-wear process provide an example for which regression model can sometimes be useful in modeling and controlling the process. For the control of such process, we present a simple method for testing first order autocorrelation in regression errors. The method is based on Bayesian test method via Bayes factor and derived by observing that in general, a Bayes factor can be written as the product of a quantity called the Savage-Dickey density ratio and a correction factor ; both terms are easily estimated from Gibbs sampling technique. Performance of the method is examined by means of Monte Carlo simulation. It is noted that the test not only achieves satisfactory power but eliminates the inconvenience occurred in using the well-known Durbin-Watson test.

• International Journal of Computer Science & Network Security
• /
• 제21권4호
• /
• pp.84-92
• /
• 2021

Care services are a significant asset in human life. Care in its overall nature focuses on human needs and covers several aspects such as health care, homes, personal care, and education. In fact, care deals with many dimensions: physical, psychological, and social interconnections. Very little information is available on estimating the cost of care services that provided to orphans and abandoned children. Prediction of the cost of the care system delivered by governmental or non-governmental organizations to support orphans and abandoned children is increasingly needed. The purpose of this study is to analyze the care cost for orphanage organizations in Saudi Arabia to forecast the cost as well as explore the most influence factor on the cost. By using business analytic process that applied statistical and machine learning techniques, we proposed a model includes simple linear regression, Naive Bayes classifier, and Random Forest algorithms. The finding of our predictive model shows that Naive Bayes has addressed the highest accuracy equals to 87% in predicting the total care cost. Our model offers predictive approach in the perspective of business analytics.

• Journal of the Korean Data and Information Science Society
• /
• 제23권4호
• /
• pp.627-641
• /
• 2012

본 연구는 다차원 분할표 형태로 정리된 범주형 자료가 결측치나 무응답을 가지고 있을 때 주어진 자료를 가장 잘 설명하고 예측의 정확도를 높일 수 있는 모형의 추정과 모형의 선택 문제를 다루었다. 무시할 수 없는 무응답 (non-ignorable non-response)체계하에서 최대우도 추정에서 발생할 수 있는 변방값 문제를 해결하기 위하여 계층적 베이지안 모형을 고려하였다. 또한 모형 적도를 높이기 위한 변수 조합을 찾는 모형 선택의 문제를 함께 다루었다. 베이지안 접근하에서 모형 선택의 문제를 다루기 위하여 베이즈 인자 (Bayes factor)를 모형 선택의 기준으로 이용하였다. 제시된 방법은 2004년 실시된 우리나라 국회의원 선거를 앞두고 수행된 여론조사 데이터를 이용하여 실증분석을 수행하였다. 분석결과 무시할 수 없는 무응답 체계하에서 설명변수로 투표참여여부를 이용하는 것이 가장 적합한 모형으로 판명되었다.

• 한국전기전자재료학회논문지
• /
• 제30권9호
• /
• pp.551-557
• /
• 2017

It is extremely important to improve methodologies for the lifetime assessment of porcelain insulators. While there has been a considerable amount of work regarding the phenomena of lifetime distributions, most of the studies assume that aging distributions follow the Weibull distribution. However, the true underlying distribution is unknown, giving rise to unrealistic inferences, such as parameter estimations. In this article, we review several distributions that are commonly used in reliability and survival analysis, such as the exponential, Weibull, log-normal, and gamma distributions. Some properties, including the characteristics of failure rates of these distributions, are presented. We use a Bayesian approach for model selection and parameter estimation procedures. A well-known measure, called the Bayes factor, is used to find the most plausible model among several contending models. The posterior mean can be used as a parameter estimate for unknown parameters, once a model with the highest posterior probability is selected. Extensive simulation studies are performed to demonstrate our methodologies.

• 응용통계연구
• /
• 제24권6호
• /
• pp.1149-1160
• /
• 2011

기존의 frequentist 추론에 비해 Bayesian 추론에서의 가설 검정 및 모형 선택 문제는 학자들 간에 일치된 견해를 보이지 못하고 있으며 아직도 논란이 되는 것들이 많다. Bayesian 추론에서 가설 검정 및 모형 선택의 기준으로 널리 쓰이는 Bayes factor는 이해하기 쉬우나 여러 경우에 구하기 어려운 단점이 존재한다. 그 외에 다른 기준으로 Spiegelhalter 등 (2002)가 제시한 DIC(Deviance Information Criterion)과 frequentist 추론에서의 P-value에 대비되는 Bayesian P-value가 있다. 본 논문에서는 Swiss banknote 자료를 Bayesian 로지스틱 회귀모형으로 분석하고 관련 기준들을 구하여 각 기준들이 일관성 있는 결론을 보이는지 확인하고자 한다.