• 응용통계연구
• /
• 제28권1호
• /
• pp.61-74
• /
• 2015

최근 교육학, 경영학, 심리학 등 사회과학 뿐만 아니라 공정관리, 생물정보학 등 자연과학에서도 널리 사용되고 있는 구조식 모형(structural equation modeling)에서 잠재변수점수(latent variable score)는 직접 측정이 불가능한 잠재변수를 수량화한 추정치이다. 이 연구에서는 구조식 모형을 단계(stage)별로 분할하여 분석하는 단계별 구조식 모형(stagewise SEM; SSEM)을 제안하였다. 기존 방법은 모든 관측변수의 분산-공분산을 한꺼번에 고려하므로 독립변수인 외생잠재변수(exogenous latent variable)가 종속변수인 내생잠재변수(endogenous latent variable)에 의해 영향을 받는, 논리적으로 타당하지 않은 경우가 있다. 단계별 구조식 모형은 이런 문제점을 해결할 뿐만 아니라 모형의 복잡성을 낮추어 쉽게 해를 찾을 수 있으며, 분석과정에서 생성되는 잠재변수점수로 추가 분석도 용이하다.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2005년도 ICCAS
• /
• pp.2516-2520
• /
• 2005

In this paper, finite impulse response (FIR) smoothers are proposed for discrete-time systems. The proposed FIR smoother is designed under the constraints of linearity, unbiasedness, FIR structure, and independence of the initial state information. It is also obtained by directly minimizing the performance criterion with unbiased constraints. The approach to the MVF smoother proposed in this paper is logical and systematic, while existing results have heuristic assumption, such as infinite covariance of the initial state. Additionally, the proposed MVF smoother is based on the general system model that may have the singular system matrix and has both system and measurement noises. Thorough simulation studies, it is shown that the proposed MVF smoother is more robust against modeling uncertainties numerical errors than fixed-lag Kalman smoother which is infinite impulse response (IIR) type estimator.

• 음성과학
• /
• 제12권2호
• /
• pp.171-182
• /
• 2005

An efficient GMM (Gaussian mixture modeling) method based on LPCA (local principal component analysis) with VQ (vector quantization) for speaker identification is proposed. To reduce the dimension and correlation of the feature vector, this paper proposes a speaker identification method based on principal component analysis. The proposed method firstly partitions the data space into several disjoint regions by VQ, and then performs PCA in each region. Finally, the GMM for the speaker is obtained from the transformed feature vectors in each region. Compared to the conventional GMM method with diagonal covariance matrix, the proposed method requires less storage and complexity while maintaining the same performance requires less storage and shows faster results.

• 산업경영시스템학회지
• /
• 제19권38호
• /
• pp.117-129
• /
• 1996

This Study intends to analysis what 3 factors, which are indices of Capital, Labor and Distribution, really affect to Value-Added Productivity through Statistical Analysis. For this, We selected 12 indices of Value-Added from the edition of 'Annual report of Korean companies' published in 'Korea Investors Service., Inc', especially in parts of Chemicals and Chemical products of total 85 companies. Using this data, Multivariate Statistical Analysis such as Principal Component Analysis, Factor Analysis, Covariance Structure Analysis is taken for modeling the effect of 3 factor(Labor Productivity, Capital Productivity and the Index of Distribution) on Value-Added Productivity.

• 아동학회지
• /
• 제22권2호
• /
• pp.113-132
• /
• 2001

In order to develop a model for better understanding of causal relationships in resiliency factors in families of children with disabilities, 200 families participated in this adaptation of the Resiliency Model of McCubbin and McCubbin(1993). The 6 latent variables included in the hypothesized model were family stress, family hardiness, family schema, community support, family problem-solving communication, and family adaptation. The models were developed on the basis of confirmatory factor analysis and compared using covariance structure modeling (LISREL). Adequate fitness of the model was observed. Family stress showed negative effect on family schema and on family hardiness. Family schema showed positive effect on community support and on family hardiness. Family hardiness showed positive effect on family problem-solving communication, and family problem-solving communication showed positive effect on family adaptation.

• 한국정보시스템학회지:정보시스템연구
• /
• 제14권1호
• /
• pp.227-249
• /
• 2005

In this research, four antecedents of trust ad the effect of trust and perceived risk on attitude and purchase intension in internet shopping malls is investigated. In survey, data were collected from 286 male and female internet shopping mall users, and then covariance structure modeling through Amos was used to test the 10 hypotheses. The survey results as follow: First, this research shows that institution base structural assurance and situational normality and knowledge-based familiarity are important antecedents on trust in internet shopping malls. Second, trust affected attitude and purchase intension. Third, it has a negative relationship between trust and perceived risk. But perceived risk does not influence attitude and purchase intension. This study contribute to understanding on the role of trust and perceived risk in internet shopping malls.

• Asian-Australasian Journal of Animal Sciences
• /
• 제15권8호
• /
• pp.1222-1226
• /
• 2002

Estimates of genetic correlation between direct and maternal effects for weaning weight of beef cattle are often negative in field data. The biological existence of this genetic antagonism has been the point at issue. Some researchers perceived such negative estimate to be an artifact from poor modeling. Recent studies on sources affecting the genetic correlation estimates are reviewed in this article. They focus on heterogeneity of the correlation by sex, selection bias caused from selective reporting, selection bias caused from splitting data by sex, sire by year interaction variance, and sire misidentification and inbreeding depression as factors contributing sire by year interaction variance. A biological justification of the genetic antagonism is also discussed. It is proposed to include the direct-maternal genetic covariance in the analytical models.

• 대한전기학회논문지
• /
• 제39권11호
• /
• pp.1207-1218
• /
• 1990

The lack of persistent excitation (PE) can be the reason of freezing in the recursive least square estimators and the covariance windup in the exponential weighted least square estimators. We present a theoretical analysis of these phenomena and a simple method to check the exciting condition in real time. Using these results and under some conditions such as slowly time varying Plant and a tracking problem for set point, a robust self-tuning regulators without PE is proposed. In this algorithm, when PE is not satisfied, only plant gain is estimated, and then the system parameters are corrected by it. It is shown that the gain adaptive scheme makes the robustness to be improved against modeling error, off-set, and correlated noise etc, by the results of analysis and simulations.

• Journal of the Korean Data and Information Science Society
• /
• 제26권2호
• /
• pp.495-504
• /
• 2015

The Gaussian geostatistical model has been widely used for modeling spatial data. However, this model suffers from a severe difficulty in computation because inference requires to invert a large covariance matrix in evaluating log-likelihood. In addressing this computational challenge, three strategies have been employed: likelihood approximation, lower dimensional space approximation, and Markov random field approximation. In this paper, we reviewed statistical approaches attacking the computational challenge. As an illustration, we also applied integrated nested Laplace approximation (INLA) technology, one of Markov approximation approach, to real data to provide an example of its use in practice dealing with large spatial data.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1999년도 하계학술대회 논문집 A
• /
• pp.200-202
• /
• 1999

A method of stochastic finite element analysis is developed for yield a uncertainty of engineering problems. Where, a stochastic finite-element method for shapes modeling is proposed a6 a means to solve the models with the uncertainty and variety. This method is based on the probability and illustrated by a first-Order Second-Moment Method and considering the covariance of random variables. The validity and accuracy of the stochastic finite element method is verified through comparing with those solved by the conventional 2-D finite element method.