• Journal of the Korean Data and Information Science Society
• /
• 제27권2호
• /
• pp.381-394
• /
• 2016

Graphical model represents conditional dependencies between variables as a graph with nodes and edges. It is widely used in various fields including physics, economics, and biology to describe complex association. Conditional dependencies can be estimated from a inverse covariance matrix, where zero off-diagonal elements denote conditional independence of corresponding variables. This paper proposes a efficient BCDR (block coordinate descent with random permutation) algorithm using graphics processing units and random permutation for the CONCORD (convex correlation selection method) based on the BCD (block coordinate descent) algorithm, which estimates a inverse covariance matrix based on pseudo-likelihood. We conduct numerical studies for two network structures to demonstrate the efficiency of the proposed algorithm for the CONCORD in terms of computation times.

• The Korean Journal of Applied Statistics
• /
• 제21권2호
• /
• pp.291-299
• /
• 2008

This paper discusses model building for repeated binary response data with different time-dependent covariates each occasion. Since repeated measurements data are having correlated structure, weighed least squares(WLS) methodology is applied. Repeated measures designs are usually having different sizes of experimental units like split-plot designs. However repeated measures designs differ from split-plot designs in that the levels of one or more factors cannot be randomly assigned to one or more of the sizes of experimental units in the experiment. In this case, the levels of time cannot be assigned at random to the time intervals. Because of this nonrandom assignment, the errors corresponding to the respective experimental units may have a covariance matrix. So, the estimates of effects included in a suggested logit model are obtained by using covariance structures.

• The Korean Journal of Applied Statistics
• /
• 제28권2호
• /
• pp.151-158
• /
• 2015

This paper discusses how to use projections for the analysis of data from balanced incomplete block designs. A model is suggested as a matrix form for the interblock analysis. A second set of treatment effects can be found by projections from the suggested interblock model. The variance and covariance matrix of two estimated vectors of treatment effects is derived. The uncorrelation of two estimated vectors can be verified from their covaraince structure. The fitting constants method is employed for the calculation of block sum of squares adjusted for treatment effects.

• Journal of Animal Science and Technology
• /
• 제47권5호
• /
• pp.703-710
• /
• 2005

Data from Hanwoo steers and bull calves were analyzed to see the phenotypic and genetic relationships between carcass traits from four different covariance models. Four models fit test station and test period as fixed effect of contemporary group and sire as random effect assuming paternal half-sib relationships among animals. Each model fits one of linear covariate (s) of different slaughter end points-age at slaughter in the first order, age at slaughter in the first and second order, slaughter weight or back fat thickness at 12-13th rib of cold carcass. Age at slaughter in its second order was not significant. Age at slaughter accounted for signifi- cant amount of genetic variances and covariances of carcass traits. Heritability estimates of back fat thickness, rib eye area, carcass weight, marbling score and dressing percentage were 0.34, 0.22, 0.24, 0.42 and 0.18, respectively at constant age basis. The genetic correlation between carcass weight and the other variables were all positive and low to high in magnitude. Genetic correlations between back fat thickness and rib eye area and between marbling score and dressing percentage were low but negative. Variance and covariance structure between these traits were shifted to a great extent when these variables were regressed on slaughter weight or on back fat thickness. These two covariates counteracted to each other but they adjusted each carcass variable or their interrelationship according to differential growth of body components, bone, muscle and fat. Slaughter weight tended to decrease genetic variances and covariances of carcass weight and between component traits and back fat thickness tended to increase those of rib eye area and between rib eye area and carcass weight.

• Journal of the Korean Data and Information Science Society
• /
• 제25권5호
• /
• pp.1025-1038
• /
• 2014

Linear mixed model has often been utilized for genetic association analysis with family-based samples. The correlation matrix for family-based samples is constructed with kinship coefficient and assumes that parental phenotypes are independent and the amount of correlations between parent and offspring is same as that of correlations between siblings. However, for instance, there are positive correlations between parental heights, which indicates that the assumption for correlation matrix is often violated. The statistical validity and power are affected by the appropriateness of assumed variance covariance matrix, and in this thesis, we provide the linear mixed model with flexible variance covariance matrix. Our results show that the proposed method is usually more efficient than existing approaches, and its application to genome-wide association study of body mass index illustrates the practical value in real data analysis.

• The Korean Journal of Applied Statistics
• /
• 제1권1호
• /
• pp.39-44
• /
• 1987

For obtaining the m.l.e. of $\Sigma$ from p-variate Non-singular Normal parent, $N_p(\mu, \Sigma)$, Andersous' Procedure based on the invariance property of the m.l.e. seems to be generally Preferred in the view of its simplicity. This paper shows that his approach with respect to $\Sigma^{-1}$ rather than $\Sigma$ itself, he burther applicable to deriving the m.l.e. of parametersinvolved in the common factor model an dsimplex model as well.

• Proceedings of the Korean Association for Survey Research Conference
• /
• 한국조사연구학회 2007년도 추계학술대회 발표논문집
• /
• pp.55-65
• /
• 2007

In this research, it will be examined on mathematical model of AMOS software program that ues for Covariance Structure Analysis. if we have not understood to mathematical model of Covariance Structure, we fail to understand Structural equation modeling. Similarly If We were not understand to mathematical model of AMOS Software, we do not use Software adequately. Therefore we examine two sorts of Software that be designed for Structural equation modeling or Covariance Structure Analysis. In this research, We will focus on 8 assumption of Structural equation modeling and compare AMOS(Analysis of MOment Structure) program with LISREL(Linear Structure RELation) program. We found that A Program of AMOS Software have materialized with RAM(Reticular Action Model).

• The Korean Journal of Applied Statistics
• /
• 제21권4호
• /
• pp.621-630
• /
• 2008

This paper discusses how to construct a generalized marginal logit model for analyzing repeated polytomous response data when some factors are applied to larger experimental units as treatments and time to a smaller experimental unit as a repeated measures factor. So, two different experimental sizes are considered. Weighted least squares(WLS) methods are used for estimating fixed effects in the suggested model.

• Proceedings of the Earthquake Engineering Society of Korea Conference
• /
• 한국지진공학회 2002년도 추계 학술발표회 논문집
• /
• pp.213-220
• /
• 2002

본 연구에서는 최근에 개발된 Unscented Particle Filter (UPF)를 사용한 비선형 동적 구조계의 구조계수 규명기법이 연구되었다. 일반적인 비선형 구조계수 추정 문제의 일반 해는 존재하지 않으나, 그에 대한 대안으로써 선형 근사 기법인 extended Kalman filter (EKF)가 비선형 동적 구조계수의 추정에 주로 사용되어왔다. 그러나, EKF는 구간 선형(piecewise linear) 가정으로 인해 biased estimator이고 비선형성이 상대적으로 높을 때 오차가 큰 추정치를 주는 단점을 가진다. 이를 보완하기 위해서 UPF가 개발되었고, 이 기법은 particle filter의 일종으로써 Unscented Kalman filter (UKF)를 사용하여 importance proposal distribution을 생성한다. 수치실험이 SDOF와 MDOF에 대하여 3가지 경우에 대해서 수행되었다. 비선형 SDOF의 수치 실험으로부터 잡음이 가해진 상태에서 UKF가 EKF에 비해 초기 공분산 행렬의 가정에 대해 정확하고 강인한 추정결과를 보여줌을 보였다 최하층의 column에 비선형 거동이 발생하는 5층 전단 빌딩모형의 수치실험으로부터 UKF가 복잡한 구조물의 구조계수 추정능력이 있음을 보여주었다. 여러 가지 수치실험은 UPF가 EKF보다 비선형 동적 구조계수 추정에 있어서 더 나은 방법임을 보여 주었다.

• The Korean Journal of Applied Statistics
• /
• 제22권6호
• /
• pp.1265-1275
• /
• 2009

In this paper, we investigated the composite covariance structure models for repeated measures data with multiple repeat factors. When the number of repeat factors is more than three, it is infeasible to fit the composite covariance models using the existing statistical packages. In order to fit the composite covariance structure models to real data, we proposed two approaches: the dimension reduction approach for repeat factors and the random effect model approximation approach. Our proposed approaches were illustrated by using the blood pressure data with three repeat factors obtained from 883 subjects.