• Asian-Australasian Journal of Animal Sciences
• /
• 제11권6호
• /
• pp.642-647
• /
• 1998

A Poisson error model as a generalized linear mixed model (GLMM) has been suggested for genetic analysis of counted observations. One of the assumptions in this model is the normality for random effects. Since this assumption is not always appropriate, a more flexible model is needed. For count traits, a Poisson hierarchical generalized linear model (HGLM) that does not require the normality for random effects was proposed. In this paper, a Poisson-Gamma HGLM was examined along with corresponding analytical methods. While a difficulty arises with Poisson GLMM in making inferences to the expected values of observations, it can be avoided with the Poisson-Gamma HGLM. A numerical example with simulated embryo yield data is presented.

• Journal of the Korean Data and Information Science Society
• /
• 제27권4호
• /
• pp.1083-1090
• /
• 2016

In clinical studies, different types of outcomes (e.g. repeated measures data and time-to-event data) for the same subject tend to be observed, and these data can be correlated. For example, a response variable of interest can be measured repeatedly over time on the same subject and at the same time, an event time representing a terminating event is also obtained. Joint modelling using a shared random effect is useful for analyzing these data. Inferences based on marginal likelihood may involve the evaluation of analytically intractable integrations over the random-effect distributions. In this paper we propose a joint HGLM approach for analyzing such outcomes using the HGLM (hierarchical generalized linear model) method based on h-likelihood (i.e. hierarchical likelihood), which avoids these integration itself. The proposed method has been demonstrated using various numerical studies.

• 한국병원경영학회지
• /
• 제6권2호
• /
• pp.22-36
• /
• 2001

The hospital bankruptcy rate is increasing, therefore it is very important to predict the bankruptcy using the existing hospital management information. The hospital bankruptcy is often measured in year intervals, called grouped duration data, not by the continuous time elapsed to the bankruptcy. This study introduces a hierarchical generalized linear model(HGLM) for analysis of hospital bankruptcy data. The hazard function for each hospital may be influenced by unobservable latent variables, and these unknown variables are usually termed as random effects or frailties which explain correlations among repeated measures of the same hospital and describe individual heterogeneities of hospitals. Practically, the data of twenty bankrupt and sixty profitable hospitals were collected for five years, and were fitted to HGLM. The results were compared with those of the logit model. While the logit model resulted only in the effects of explanatory variables on the bankruptcy status at specific period, the HGLM showed variables with significant effects over all observed years. It is concluded that the HGLM with a fixed ratio and a period of total asset turnrounds was justified, and could find significant within and between hospital variations.

• Journal of the Korean Data and Information Science Society
• /
• 제19권4호
• /
• pp.1429-1440
• /
• 2008

In a meta-analysis combining the results from different clinical trials, it is important to consider the possible heterogeneity in outcomes between trials. Such variations can be regarded as random effects. Thus, random-effect models such as HGLMs (hierarchical generalized linear models) are very useful. In this paper, we propose a HGLM framework for analyzing the binominal response data which may have variations in the odds-ratios between clinical trials. We also present the prediction intervals for random effects which are in practice useful to investigate the heterogeneity of the trial effects. The proposed method is illustrated with a real-data set on 22 trials about respiratory tract infections. We further demonstrate that an appropriate HGLM can be confirmed via model-selection criteria.

• 응용통계연구
• /
• 제17권3호
• /
• pp.431-443
• /
• 2004

본 연구에서는 질병지도작성(disease mapping)을 위해 인접지역의 정보를 효과적으로 활용할 수 있는 EB(empirical Bayes) 추정 법과 HGLM(hierarchial generalized linear model)을 기초로 한 추정법을 다룬다. 사례연구로 이 추정방법들을 이용하여 2000년 사망원인통계자료를 이용해 경상도 및 전라도의 112개 시$.$군$.$구 단위 행정자치구역별 45세 이상 폐암 사망률을 산출하고, 경상도 및 전라도 지역 폐암 사망률 지도를 작성한다. 아울러 제시된 방법들에 위해 얻어진 추정치들의 변동과 3년간 평균 사망률을 기준으로 구한 MSD(mean square deviation)를 이용하여 추정방법들의 특성을 비교 분석한다.

• Journal of the Korean Data and Information Science Society
• /
• 제27권5호
• /
• pp.1389-1397
• /
• 2016

Semi-parametric frailty model with nonparametric baseline hazards has been widely used for the analyses of clustered survival-time data. The frailty models can be fitted via an auxiliary Poisson hierarchical generalized linear model (HGLM). For the inferences of the frailty model marginal likelihood, which gives MLE, is often used. The marginal likelihood is usually obtained by integrating out random effects, but it often requires an intractable integration. In this paper, we propose to obtain the MLE via Laplace approximation using a Poisson HGLM approach for semi-parametric frailty model. The proposed HGLM approach uses hierarchical-likelihood (h-likelihood), which avoids integration itself. The proposed method is illustrated using a numerical study.

• 한국사회복지학
• /
• 제58권2호
• /
• pp.119-141
• /
• 2006

본 연구의 목적은 우리나라 근로빈곤층의 실태와 특성을 파악하고 근로빈곤층의 정태적 결정요인을 파악하는데 있다. 이를 위해 본 연구는 한국노동패널조사의 제2차년도(1999년)부터 제7차년도(2004년)의 반복측정 자료를 개인간(between-person), 개인내(within-person) 2층(two-level)으로 병합하여 자료를 구성하고 이를 통해 각 수준의 변수들이 근로자의 빈곤지위여부에 미치는 영양을 위계적 일반화 선형모형(HGLM: hierarchical generalized linear model)을 이용하여 추정하였다. 분석의 결과, 우리나라 취업자 가운데 가구소득이 빈곤선 이하의 생활을 하는 근로빈곤층(개인)의 규모는 약 10.0% 내외의 규모를 보이는 것으로 나타났다. 이러한 근로계층의 빈곤지위에 영양을 미치는 요인으로는 성별, 교육수준, 결혼상태, 취업형태, 고용업종, 고용직종 등으로 밝혀졌으며 이외 가구원수, 연령 등은 유의미안 영향을 미치지 않은 것으로 나타났다.

• Journal of the Korean Data and Information Science Society
• /
• 제26권1호
• /
• pp.101-109
• /
• 2015

행정구역상 읍/면/동 단위의 소지역 (small area)별로 질병위험의 차이에 대한 분석을 위해, 2005년 기준 서울 행정동을 기준으로 2005년부터 2008년까지 질병, 사고, 암 사망자료에 대한 표준화 사망률 (SMR; standardized mortality rate)을 고려하였다. 소지역 단위로 질병사망률을 직접 추정하는 것은 소지역 내 표본수가 작아, 개발 소지역 단위에서의 직접 계산된 SMR은 그 추정치의 정도 (precision) 확보가 어려운 문제점이 발생한다. 따라서, 본 연구에서는 각 소지역간 효과 추정을 위해 공간적 상관성 (spatial correlation)을 가지는 다단계 일반화 선형모형 (HGLM; hierarchical generalized linear models)을 고려하였다. 이를 통해, 서울지역 동별 주요 사망원인에 따른 공변량의 효과 및 추정된 SMR을 근거로 질병지도 결과를 제시하였다.

• 농촌계획
• /
• 제14권3호
• /
• pp.63-73
• /
• 2008

The purpose of this study is to analyze the determinants of the digital divide at individual level and regional level in Korea, considering interaction between individual and the regional variables. Following results are obtained. First, individual level digital devide in the 16 different regions has been found in terms of Internet use, implying the needs for further analysis on impact of the regional factor in individual Internet use. Second, the result finds the impact of level-l individual variables, "gender, age, education, income and jobs" on digital divide, significantly at level 10% level. Third, the regional variables influencing the individual digital divide were not found at state level. However, regional factors might affect digital devide at county level. Study suggest some plans to reduce digital divide. First, the digital devide at individual level should be remedied by focusing on neglected class of people. Second, we need to approach the digital divide by analyzing in more detail, reflecting interactions of the regional variables and individual variables. Third, we should come up with a policy for mending the digital divide at regional level.

• Journal of the Korean Data and Information Science Society
• /
• 제22권2호
• /
• pp.137-148
• /
• 2011

본 논문에서는 최근 중요한 문제로 대두되고 있는 손상으로 인한 사망 중 운수사고, 자살, 낙상사고에 의한 사망률에 대한 시 군 구 별 차이를 체계적으로 파악하고자 한다. 2008년 사망원인통계 원시 자료 중 19세 이상이면서, 국제사인분류에 따른 사인이 운수사고, 자살, 낙상사고에 의한 자료만을 추출하여 분석대상으로 고려하였다. 분석모형으로 성별, 연령, 1인당 주민세를 고정효과로 보정하고, 사망자수가 포아송분포를 따른다는 가정 하에 지역효과를 변량효과로 둔 포아송 HGLM 모형을 고려하여 시 군 구 소지역별 효과의 차이를 질병지도로 나타내었다. 분석결과 운수사고, 자살사고로 인한 사망률은 시 군 구 소지역별로 유의한 차이가 나타났지만, 낙상사고로 인한 사망률은 시 군 구 지역별로 유의한 차이가 없는 것으로 나타났다.