• Communications for Statistical Applications and Methods
• /
• 제26권4호
• /
• pp.371-383
• /
• 2019

Panel data sets have been developed in various areas, and many recent studies have analyzed panel, or longitudinal data sets. Maximum likelihood (ML) may be the most common statistical method for analyzing panel data models; however, the inference based on the ML estimate will have an inflated Type I error because the ML method tends to give a downwardly biased estimate of variance components when the sample size is small. The under estimation could be severe when data is incomplete. This paper proposes the restricted maximum likelihood (REML) method for a random effects panel data model with a censored dependent variable. Note that the likelihood function of the model is complex in that it includes a multidimensional integral. Many authors proposed to use integral approximation methods for the computation of likelihood function; however, it is well known that integral approximation methods are inadequate for high dimensional integrals in practice. This paper introduces to use the moments of truncated multivariate normal random vector for the calculation of multidimensional integral. In addition, a proper asymptotic standard error of REML estimate is given.

• Journal of the Korean Statistical Society
• /
• 제26권2호
• /
• pp.181-198
• /
• 1997

The variable sampling rates scheme is proposed by taking random sample size and sampling interval during the process. The performance of the scheme is measured in terms of the average time to signal adjusted by teh sampling cost when the process is out of control. This measurement evaluates the effectiveness of the scheme in terms of the cost incurred due to nonconformation as well as sampling. The variable sampling rates scheme is shown to be effective especially for small and moderate shifts of the mean when compared to the standard scheme.

• 품질경영학회지
• /
• 제16권2호
• /
• pp.82-91
• /
• 1988

This study is concerned with the development of the economic sampling inspection plan when it is possible to carry out a nondestructive screening inspection for the rejected lots by substitutive characteristics closely related to the original quality characteristic. It is assumed that the probabilities of those errors depend linearly on the fraction defective of the process. The inspection policies considered are as follows: take a random sample of size n from the lot of size N and perform a destructive test on the sample. If the number of the defective items is not greater than a critical number C, accept the lot. Expected cost functions are obtained when the process average can be considered to follow a beta distribution and the way of finding the optimal values of (n, c) is to be explored.

• 한국산림과학회지
• /
• 제35권1호
• /
• pp.39-46
• /
• 1977

본연구(本硏究)는 1974년(年)에 선발(選拔)하여 그후에 수년간(數年間)에 걸처 분석(分析)한 주왕산(周王山), 안면도(安眠島), 오대산(五臺山)의 소나무(Pinus densifiora S. et Z.) 천연집단(天然集團)의 형질(形質)중에시 차대묘목(次代苗木)(1-0묘(苗), 1-1묘(苗))의 묘고(苗高) 및 근원경(根元徑)의 집단간(集團間), 가계간(家系間)의 차이(差異)를 추정(推定)함에 있어 가계(家系)의 수(數)를 어느 정도의 크기로 해야만 좋은 추정(推定)을 할 수 있는 가를 알아보기 위함이었다. 본 분석(分析)은 Nested Design의 2-level의 equal sample size에 의한 방법(方法)과 Randomized Complete Block Design에 의한 방법(方法) 두 가지를 이용하였으며 그 결과(結果)를 요약(要約)하면 다음과 같다. 1. Nested design에 의한 분석(分析)에 있어서 sample size(가계수(家系數))의 증가(增加)에 따라 가계간(家系間)의 F-치(値)는 대 체로 감소(減少)하나 집단간(集團間)의 F-치(値)는 점점 증가(增加)하는 경향(傾向)을 보였다. 2. F-치(値)의 증가(增加) 및 감소(減少)의 경향(傾向)은 그 sample size가 적을 때는 크게 변화(變化)를 보이나 그의 크기가 대략 10을 넘으면서부터는 다소 안정(安定)된 값을 보이고 있어 만일 sample size가 15~20정도가 되면 상당히 안정되고 그에 따라 집단간차이(集團間差異)를 추정(推定)하는데 오류(誤謬)를 범할 확률(確率)은 대단히 낮아질 것으로 추론(推論)되었다. 3. 연속적(連續的)으로 누적가산(累積加算)된 sample size를 가지고 분석(分析)할 때와 추출조합(抽出組合)된 sample을 가지고 분석(分析)할 때 F-치(値)의 변동(變動)은 후자(後者)가 훨씬 심하고 또 sample size가 적을때 그런 차이(差異)가 심해지는 것을 볼 수 있었다. 또한 분석대상형질(分析對象形質)이 다른 때에는 역시 그에 따른 분석가계수(分析家系數)의 조정(調整)이 필요(必要)할 것으로 보였다. 4. Randomized Complete Block Design에 의한 분석(分析)에서 가계간(家系間)의 차이(差異)에 대(對)한 분석(分析)은 집단간(集團間)의 추정(推定)에 필요(必要)한 가계수(家系數)보다는 적은 가계수(家系數)로도 분석가능(分析可能)하였다. 이것은 일반적(一般的) 추리(推理)를 그대로 입증(立證)해 주는 것에 불과(不過)하였다. 반복구간(反復區間)의 차이(差異)에 대한 추정(推定)에서 Sample Size가계수(家系數)의 영향(影響)은 일정하게 말하기 힘들었다.

• Journal of Nutrition and Health
• /
• 제17권3호
• /
• pp.236-241
• /
• 1984

In order to make clear the relationship between sample design and sample survey in community, it was conducted research on sample design for National Nutrition Survey in 1983. In this paper it was tried to analize the data based on The Report of a Settled Population, 1981 conducted by National Bureau of Statistics Economic Planning Board. The sample was basically using stratified two-stage sampling with systematic sampling of Ban or Li as administrative unit. The population represents the whole nation excluding Jeju-do because of budget. The selection of sampling unit and sampling procedure was as follows. 1) Stratify the nation-wide area in 20 sections according to administrative districts. 2) Determine the sample size in each section according to equal proportional rate (1 / 8040) and to about 1,000 households in the sample. 3) Select the 25 sampling units by section according to households proportion. 4) Select the 10 households at random from each Ban or Li according to equal probability proportion as the final sampling unit. Using the procedure, it was sampled 1,000 households for National Nutrition Survey in 1983.

• 응용통계연구
• /
• 제10권2호
• /
• pp.339-352
• /
• 1997

소표본으로부터 형성되는 2원 분할표에는 빈도가 작은 칸들이 적지 않기 때문에 대표본이론에 근거한 카이제곱 검증 등 기존 통계적 방법의 적용이 적절하지 않응 수가 있다. 이런 경우에 한 대안으로서 정확검증법(exact tests)이 개발되어 있으나 이것이 너무 많은 계산을 요구하므로 사용하기가 쉽지 않다. 본 연구는 소표본 2원 분할표에서, 단순한 몬테칼로 알로리즘에 의한 행 균일성 가설의 카이제곱 임의화 검증법(randomization test)을 제안하고 튜키(Tukey) 형의 행간 다중비교법을 제안한다. 아울러 열 범주가 순서형인 2원 분할표에 대하여도 유사한 방법론을 적용할 수 있음을 밝힌다.

• Journal of applied mathematics & informatics
• /
• 제10권1_2호
• /
• pp.1-15
• /
• 2002

In this paper an important stability property of the extremes under power normalizations is discussed. It is proved that the restricted convergence of the Power normalized extremes on an arbitrary nondegenerate interval implies the weak convergence. Moreover, this implication, in an important practical situation, is obtained when the sample size is considered as a random variable distributed geometrically with mean n.

• 품질경영학회지
• /
• 제19권2호
• /
• pp.86-95
• /
• 1991

Empirical Bayes estimation of the binomial parameter ${\theta}$ is considered when the sample size in each component problem is random. The result is applied to the lot acceptance sampling problem in the presence of various kinds of costs including cost for sampling.

• Journal of the Korean Data and Information Science Society
• /
• 제16권3호
• /
• pp.705-712
• /
• 2005

Entropy is the basic concept of information theory. It is well defined for random varibles with known probability density function(pdf). For given data with unknown pdf, entropy should be estimated. Usually, estimation of entropy is based on the approximations. In this paper, we consider a kernel based approximation and compare it to the cumulant approximation method for several distributions. Monte carlo simulation for various sample size is conducted.

• 한국의류학회지
• /
• 제2권2호
• /
• pp.227-235
• /
• 1978

This study was designed to find out consumer's interest in garments label and to help home economists make interest for the further study in relation between producers and consumers as gap bridger. The questionnair method was used to obtained the data which was made by a result of self-administered questionnair. A size of random sample for this research was 364 subjects. The study found the following: (1) Most of consumers are relatively interested in garments label. The most concious age level was woman of fourty. (2) The most interest factor was label of size, price, fiber contents. brand name, directions and precautions on proper use and care. (3) The order of complaining item after washing was change of size, and color, seam pucker. deformation of collar. and button. (4) Most of consumers do not follow the direction when they clean their garments. (5) The respondents seem to be not understand the garment's informative label.