• 융합정보논문지
• /
• 제12권5호
• /
• pp.100-106
• /
• 2022

종단연구는 동일 대상에 대하여 반복적으로 측정한 종단자료를 기반으로 하는 연구방법을 말한다. 대부분의 종단분석 방법은 예측이나 추론에 적합하고, 탐색적 목적으로 사용하기에는 적합하지 않은 경우가 많다. 본 연구에서는 종단자료를 분석하는 탐색적 방법을 제시한다. 이 방법은 자기조직화지도기법을 사용하여 종단자료를 군집화 하여 최선의 군집 수를 정한 후 종단궤적을 찾는 방법이다. 제안한 방법론은 고용정보원의 종단자료에 적용되었으며, 총 2,610개의 샘플에 대하여 분석을 하였다. 방법론을 적용한 결과 패널 별로 시계열적으로 군집 화되는 결과를 얻었다. 이는 종단자료를 사전에 클러스터링하고 다층 종단분석을 하는 것이 더욱 효과적이라는 사실을 나타낸다.

• Journal of the Korean Data and Information Science Society
• /
• 제21권2호
• /
• pp.363-369
• /
• 2010

In this paper we propose a mixed-effects least squares support vector regression (LS-SVR) for longitudinal data. We add a random-effect term in the optimization function of LS-SVR to take random effects into LS-SVR for analyzing longitudinal data. We also present the model selection method that employs generalized cross validation function for choosing the hyper-parameters which affect the performance of the mixed-effects LS-SVR. A simulated example is provided to indicate the usefulness of mixed-effect method for analyzing longitudinal data.

• 응용통계연구
• /
• 제29권7호
• /
• pp.1459-1473
• /
• 2016

시간의 흐름에 따라 관측되는 경시적(longitudinal) 자료의 경우, 경시적 자료와 생존(survival) 자료가 종종 동시에 수집된다. 이 때 경시적 자료에서 발생하는 결측이 생존자료와의 연관성으로 인해 발생한 무시할 수 없는 결측(non-ignorable missing)이라면, 경시적 자료분석 방법만으로는 두 자료 간의 연관성을 고려하지 않아 독립변수에 대한 효과는 편향된 결과를 얻게 된다. 이러한 문제를 해결하기 위해서 결측의 원인이 생존시간과 연관되어 있으므로 생존모형을 고려하여 불편추정량을 얻기 위해 경시적 자료와 생존자료의 결합모형에 대한 연구가 이루어져 왔다. 본 논문은 경시적 자료의 형태가 영이 많이 존재하는 영과잉 가산자료(zero-inflated count data)와 생존자료의 결합모형을 연구하였다. 경시적 영과잉 가산자료와 생존자료는 각각 허들모형(hurdle model)과 비례위험모형(proportional hazards model)의 부 모형을 적용하였고, 두 부 모형들의 변량효과가 다변량 정규분포를 따른다는 가정을 통하여 결합하였다. 모수의 최우추정법으로 EM 알고리즘을 활용하였고, 추정된 표준오차를 계산하기 위해 프로파일 우도(profile likelihood)를 이용하였다. 최종적으로 모의실험을 통해 두 부 모형의 변량효과 간 상관관계가 존재하는 경우 결합모형이 개별적 모형보다 편의와 포함확률(coverage probability)의 측면에서 더 우수함을 보였다.

• 의학교육논단
• /
• 제25권2호
• /
• pp.84-101
• /
• 2023

Longitudinal data can provide important evidence with the potential to stimulate innovation and affect policies in medical education and can serve as a driving force for further developments in medical education through evidence-based decisions. Tracking and observing cohorts of students and graduates using longitudinal data can be a way to link the past, present, and future of medical education. This study reviewed practical methods and technical, administrative, and ethical considerations for the establishment and operation of a longitudinal database and presented examples of longitudinal databases. Cohort study design methods and previous examples of research using longitudinal databases to explore major topics in medical education were also reviewed. The implications of this study are as follows: (1) a systematic design process is required to establish longitudinal data, and each university should engage in ongoing deliberation about this issue; (2) efforts are needed to alleviate "survey fatigue" among respondents and reduce the administrative burden of those conducting data collection and analysis; (3) it is necessary to regularly review issues of personal information protection, data security, and ethics regarding the survey respondents; and (4) a system should be established that integrates and manages a longitudinal database of medical education at the national level. The hope is that establishing longitudinal data and cohorts at individual medical schools will not be a temporary phenomenon, but rather that they will be well utilized at the national level to innovate and implement ongoing changes in medical education.

• 의학교육논단
• /
• 제24권2호
• /
• pp.156-171
• /
• 2022

The purpose of this study was to establish a model for constructing longitudinal data for medical school, and to structure cohort and longitudinal data using data from Yonsei University College of Medicine (YUCM) according to the established input-environment-output (I-E-O) model. The study was conducted according to the following procedure. First, the data that YUCM has collected was reviewed through data analysis and interviews with the person in charge of each questionnaire. Second, the opinions of experts on the validity of the I-E-O model were collected through the first expert consultation, and as a result, a model was established for each stage of medical education based on the I-E-O model. Finally, in order to further materialize and refine the previously established model for each stage of medical education, secondary expert consultation was conducted. As a result, the survey areas and time period for collecting longitudinal data were organized according to the model for each stage of medical education, and an example of the YUCM cohort constructed according to the established model for each stage of medical education was presented. The results derived from this study constitute a basic step toward building data from universities in longitudinal form, and if longitudinal data are actually constructed through this method, they could be used as an important basis for determining major policies or reorganizing the curricula of universities. These research results have implications in terms of the management and utilization of existing survey data, the composition of cohorts, and longitudinal studies for many medical schools that are conducting surveys in various areas targeting students, such as lecture evaluation and satisfaction surveys.

• Communications for Statistical Applications and Methods
• /
• 제28권4호
• /
• pp.369-391
• /
• 2021

The vlda is an R (R Development Core team et al., 2011) package which provides functions for visualization of multidimensional longitudinal data. In particular, the R package vlda was developed to assist in producing a plot that more effectively expresses changes over time for two different types (long format and wide format) and uses a consistent calling scheme for longitudinal data. The main features of this package allow us to identify the relationship between categories and objects using an indicator matrix with object information, as well as to cluster objects. The R package vlda can be used to understand trends in observations over time in addition to identifying relative relationships at a simple visualization level. It also offers a new interactive implementation to perform additional interpretation, therefore it is useful for longitudinal data visual analysis. Due to the synergistic relationship between the existing VLDA plot and interactive features, the user is empowered by a refined observe the visual aspects of the VLDA plot layout. Furthermore, it allows the projection of supplementary information (supplementary objects and variables) that often occurs in longitudinal data of graphs. In this study, practical examples are provided to highlight the implemented methods of real applications.

• Genomics & Informatics
• /
• 제20권2호
• /
• pp.16.1-16.12
• /
• 2022

Various methodologies for the genetic analysis of longitudinal data have been proposed and applied to data from large-scale genome-wide association studies (GWAS) to identify single nucleotide polymorphisms (SNPs) associated with traits of interest and to detect SNP-time interactions. We recently proposed a grid-based Bayesian mixed model for longitudinal genetic data and showed that our Bayesian method increased the statistical power compared to the corresponding univariate method and well detected SNP-time interactions. In this paper, we further analyze longitudinal obesity-related traits such as body mass index, hip circumference, waist circumference, and waist-hip ratio from Korea Association Resource data to evaluate the proposed Bayesian method. We first conducted GWAS analyses of cross-sectional traits and combined the results of GWAS analyses through a meta-analysis based on a trajectory model and a random-effects model. We then applied our Bayesian method to a subset of SNPs selected by meta-analysis to further discover SNPs associated with traits of interest and SNP-time interactions. The proposed Bayesian method identified several novel SNPs associated with longitudinal obesity-related traits, and almost 25% of the identified SNPs had significant p-values for SNP-time interactions.

• Journal of the Korean Data and Information Science Society
• /
• 제17권4호
• /
• pp.1053-1065
• /
• 2006

This paper proposes a method in one-way layouts for longitudinal data with informative drop-out. When dropouts are informative, that is, correlated with unobserved data and/or the previous observed data, the simple imputation methods such as 'last observation carried forward' (LOCF) methods would arise the bias of the testing models. The maximum likelihood procedure combined with a logit model for the drop-out process is proposed to test treatment effects for one factorial designs and compared with LOCF method in two examples.

• 한국수학교육학회지시리즈D:수학교육연구
• /
• 제27권1호
• /
• pp.129-150
• /
• 2024

This study utilized South Korean elementary and middle school student data to examine the longitudinal change trajectories of learning motivation types according to the longitudinal change trajectories of mathematics academic achievement. Growth mixture modeling, latent growth model, and multiple indicator latent growth model were used to examine various change trajectories for longitudinal data. As a result of the analysis, it was classified into 4 subgroups with similar longitudinal change trajectories of mathematics academic achievement, and the characteristics of the mathematics subject, which emphasize systematicity, appeared. Furthermore, higher mathematics academic achievement was associated with higher self-determination and higher academic motivation. And as the grade level increases, amotivation increases and self-determination decreases. This study suggests that teaching and learning support using this is necessary because the level of learning motivation according to self-determination is different depending on the level of mathematics academic achievement reflecting the characteristics of the student.

• Communications for Statistical Applications and Methods
• /
• 제22권6호
• /
• pp.589-598
• /
• 2015

In longitudinal studies missing data are common and require a complicated analysis. There are two popular modeling frameworks, pattern mixture model (PMM) and selection models (SM) to analyze the missing data. We focus on the PMM and we also propose Bayesian pattern mixture models using generalized linear mixed models (GLMMs) for longitudinal binary data. Sensitivity analysis is used under the missing not at random assumption.