• The Korean Journal of Applied Statistics
• /
• v.28 no.2
• /
• pp.335-342
• /
• 2015

Joint hierarchical generalized linear models proposed by Molas et al. (2013) extend the simple longitudinal model into multiple models fitted jointly. It can easily handle the correlation of multivariate longitudinal data. In this paper, we apply this method to analyze KoGES cohort dataset. Fixed unknown parameters, random effects and variance components are estimated based on a standard framework of h-likelihood theory. Furthermore, based on the conditional Akaike information criterion the correlated covariance structure of random-effect model is selected rather than an independent structure.

• Journal of Institute of Control, Robotics and Systems
• /
• v.5 no.6
• /
• pp.714-723
• /
• 1999

This paper presents a short-term scheduling algorithm for the operation of steelmaking processes. The scope of the problem covers refining of the hot iron transferred form a blast furnace, ladle treatment, continuous casting, hot-rolling, and coiling for the final products that should satisfy the given demand. The processing time at each unit depends on how much the batch amount is treated, and te dedicated intermediate storage with finite capacity between the units is considered. Resource constraints and initial amount of each state are incorporated into the presented scheduling model for the algorithm of on-line scheduling. We propose amixed integer linear programming (MILP) model with two objectives for the scheduling. The first is to maximize the total profit while atisfying the due date constraint for each product. And the second is to minimize the total processing time, makespan, while satisfying the demand for each product. Especially, we observe the effect of penalizing the intermediate storage and the inventory level of the final product on the scheduling results.

• Journal of the Korea Society of Computer and Information
• /
• v.24 no.11
• /
• pp.41-49
• /
• 2019

We are to build an unsupervised machine learning-based language model which can estimate the amount of information that are in need to process words consisting of subword-level morphemes and syllables. We are then to investigate whether the reading times of words reflecting their morphemic and syllabic structures are predicted by an information-theoretic measure such as surprisal. Specifically, the proposed Morfessor-based unsupervised machine learning model is first to be trained on the large dataset of sentences on Sejong Corpus and is then to be applied to estimate the information-theoretic measure on each word in the test data of Korean words. The reading times of the words in the test data are to be recruited from Korean Lexicon Project (KLP) Database. A comparison between the information-theoretic measures of the words in point and the corresponding reading times by using a linear mixed effect model reveals a reliable correlation between surprisal and reading time. We conclude that surprisal is positively related to the processing effort (i.e. reading time), confirming the surprisal hypothesis.

• Journal of Korean Dental Science
• /
• v.15 no.1
• /
• pp.1-8
• /
• 2022

Purpose: Intraoral scanners, desktop scanners, and cone-beam computed tomography (CBCT) are being used in a complementary way for diagnosis and treatment planning. Limited patient-based results are available about dimensional reproducibility among different three-dimensional imaging systems. This study aimed to evaluate dimensional reproducibility among patient-derived digital models created from an intraoral scanner, desktop scanner, and two CBCT systems. Materials and Methods: Twenty-nine arches from sixteen patients who were candidates for implant treatments were enrolled. Different types of CBCT systems (KCT and VCT) were used before and after the surgery. Polyvinylsiloxane impressions were taken on the enrolled arches after the healing period. Gypsum casts were fabricated and scanned with an intraoral scanner (CIOS) and desktop scanner (MDS). Four test groups of digital models, each from CIOS, MDS, KCT, and VCT, respectively, were compared to the reference gypsum cast group. For comparison of linear measurements, intercanine and intermolar widths and left and right canine to molar lengths were measured on individual gypsum cast and digital models. All measurements were triplicated, and the averages were used for statistics. Bland-Altman plots were drawn to assess the degree of agreement between each test group with the reference gypsum cast group. A linear mixed model was used to analyze the fixed effect of the test groups compared to the reference group (α=0.05). Result: The Bland-Altman plots showed that the bias of each test group was -0.07 mm for CIOS, -0.07 mm for MDS, -0.21 mm for VCT, and -0.25 mm for KCT. The linear mixed model did not show significant differences between the test and reference groups (P>0.05). Conclusion: The linear distances measured on the digital models created from CIOS, MDS, and two CBCT systems showed slightly larger than the references but clinically acceptable reproducibility for diagnosis and treatment planning.

• Asian-Australasian Journal of Animal Sciences
• /
• v.21 no.5
• /
• pp.633-639
• /
• 2008

The genetic relationship of gestation length (GL) with birth and weaning weight (BW, WW) was investigated using data collected from the Hanwoo Experiment Station, National Institute of Animal Science, RDA, Republic of Korea. Analytical mixed models including birth year‐season, sex of calf, linear and quadratic covariates of age of dam (days) and linear covariate of age at weaning (days) as fixed effects were used. Corresponding restricted maximum likelihood (REML) and Bayesian estimates of variance components and heritability were obtained with two models; Model 1 included only direct genetic effect and Model 2 included direct genetic, maternal genetic and permanent environmental effect. All the genetic parameter estimates from REML were corresponding to the Bayesian estimates. Direct heritability estimates for GL, BW, and WW were 0.48, 0.33 and 0.25 by Model 1. From Model 2, direct and maternal heritability estimates were 0.38 and 0.03 for GL, 0.14 and 0.05 for BW, and 0.08 and 0.05 for WW. Genetic correlation estimates between direct and maternal effects were 0.05 for GL, 0.59 for BW, and 0.52 for WW. Estimates of direct genetic correlation between GL and BW (WW) were 0.44 (0.21). Positive genetic correlation of GL with BW and WW imply that selection for greater BW or WW would lead to prolonged gestation length.

• Asian-Australasian Journal of Animal Sciences
• /
• v.21 no.6
• /
• pp.784-788
• /
• 2008

Studies to detect genes responsible for economic traits in farm animals have been performed using parametric linear models. A non-parametric, model-free approach using the 'expanded multifactor-dimensionality reduction (MDR) method' considering high dimensionalities of interaction effects between multiple single nucleotide polymorphisms (SNPs), was applied to identify interaction effects of SNPs responsible for carcass traits in a Hanwoo beef cattle population. Data were obtained from the Hanwoo Improvement Center, National Agricultural Cooperation Federation, Korea, and comprised 299 steers from 16 paternal half-sib proven sires that were delivered in Namwon or Daegwanryong livestock testing stations between spring of 2002 and fall of 2003. For each steer at approximately 722 days of age, the Longssimus dorsi muscle area (LMA) was measured after slaughter. Three functional SNPs (19_1, 18_4, 28_2) near the microsatellite marker ILSTS035 on BTA6, around which the QTL for meat quality were previously detected, were assessed. Application of the expanded MDR method revealed the best model with an interaction effect between the SNPs 19_1 and 28_2, while only one main effect of SNP19_1 was statistically significant for LMA (p<0.01) under a general linear mixed model. Our results suggest that the expanded MDR method better identifies interaction effects between multiple genes that are related to polygenic traits, and that the method is an alternative to the current model choices to find associations of multiple functional SNPs and/or their interaction effects with economic traits in livestock populations.

• Journal of Embryo Transfer
• /
• v.31 no.3
• /
• pp.243-247
• /
• 2016

Growth traits, such as body weight, directly influence productivity and economic efficiency in the swine industry. In this study, we estimate heritability for body weight traits usinginformation from pedigree and genome-wide single nucleotide polymorphism (SNP) chip data. Four body weight phenotypes were measured in 1,105 $F_2$ progeny from an intercross between Landrace and Jeju native black pigs. All experimental animals were subjected to genotypic analysis using PorcineSNP60K BeadChip platform, and 39,992 autosomal SNP markers filtered by quality control criteria were used to construct genomic relationship matrix for heritability estimation. Restricted maximum likelihood estimates of heritability were obtained using both genomic- and pedigree- relationship matrix in a linear mixed model. The heritability estimates using SNP information were smaller (0.36-0.55) than those which were estimated using pedigree information (0.62-0.97). To investigate effect of common environment, such as maternal effect, on heritability estimation, we included maternal effect as an additional random effect term in the linear mixed model analysis. We detected substantial proportions of phenotypic variance components were explained by maternal effect. And the heritability estimates using both pedigree and SNP information were decreased. Therefore, heritability estimates must be interpreted cautiously when there are obvious common environmental variance components.

• Journal of Korea Water Resources Association
• /
• v.35 no.5
• /
• pp.501-510
• /
• 2002

During drought or impending drought period, the reservoir operation method is required to incorporate demand-management policy rule. The objective of this study is focused to the development of demand reduction rule by incorporating hedging-effect for a single reservoir system. To improve the performance measure of the objective function and constraints, we could incorporate three risk performance criteria proposed by Hashimoto et al. (1982) by mixed-integer programming and also incorporate successive linear programming to overcome nonlinear hedging term from the previous study(Shih et al., 1994). To verify this model, this hedging rule was applied to the Daechung multi-purpose dam. As a result, we could evaluate optimal hedging parameters and monthly trigger volumes.

• Genomics & Informatics
• /
• v.20 no.3
• /
• pp.34.1-34.11
• /
• 2022

Multilevel analysis is an appropriate and powerful tool for analyzing hierarchical structure data widely applied from public health to genomic data. In practice, however, we may lose the information on multiple nesting levels in the multilevel analysis since data may fail to capture all levels of hierarchy, or the top or intermediate levels of hierarchy are ignored in the analysis. In this study, we consider a multilevel linear mixed effect model (LMM) with single imputation that can involve all data hierarchy levels in the presence of missing top or intermediate-level clusters. We evaluate and compare the performance of a multilevel LMM with single imputation with other models ignoring the data hierarchy or missing intermediate-level clusters. To this end, we applied a multilevel LMM with single imputation and other models to hierarchically structured cohort data with some intermediate levels missing and to simulated data with various cluster sizes and missing rates of intermediate-level clusters. A thorough simulation study demonstrated that an LMM with single imputation estimates fixed coefficients and variance components of a multilevel model more accurately than other models ignoring data hierarchy or missing clusters in terms of mean squared error and coverage probability. In particular, when models ignoring data hierarchy or missing clusters were applied, the variance components of random effects were overestimated. We observed similar results from the analysis of hierarchically structured cohort data.

• International Journal of Highway Engineering
• /
• v.14 no.3
• /
• pp.25-32
• /
• 2012

The thickness of anti-freezing layer has been empirically determined using the frost depth obtained from the freezing index and has not been generally considered as a structural layer in pavement design procedure. In fact, the anti-freezing layer makes a role in structural layer and enables to reduce the total thickness of pavement system. The objective of this study is to develop the statistical regression model for evaluating the structural capacity of anti-freezing layer using Falling Weight Deflectormeter(FWD) test data in asphalt pavements. The FWD testing was conducted at the embankment, cutting, and boundary area of various test sections to estimate the structural capacity of anti-freezing layer in different foundation condition. It is observed from this testing that the center deflections of pavement structure with anti-freezing layer are smaller than those without anti-freezing layer ranging from 0.4 to 82.6%. To determine the variables of statistical model, the correlation study has been conducted between various FWD deflection indexes and the anti-freezing layer thickness. It is found that the ${\Delta}BDI$(%)(${\Delta}Basin$ Damage Index(%)) is highly correlated with anti-freezing layer thickness. The ${\Delta}BDI$(%) model were developed for evaluating structural capacity of anti-freezing layer using linear mixed-effect models.