• Proceedings of the Korean Statistical Society Conference
• /
• 2001.11a
• /
• pp.115-120
• /
• 2001

라틴 하이퍼큐브 실험계획은 전산실험을 위하여 Mckay, Beckman과 Conover(1979)에 의해 고안된 방법으로 실험을 한번 시행하는데 많은 시간과 비용이 들거나 인자가 많은 실험에 효율적으로 사용할 수 있다. 하지만 이 실험계획 역시 실험영역 전체에서 골고루 배치되지 않을 가능성이 있으므로 이를 보완하려는 시도가 이루어져 왔으며, 여기서는 good lattice points(glp)와 계통추출을 응용하여 격자기반(lattice-based) Lhd의 두 가지 방법을 제안하였다. 모의실험 결과 glp 실험계획을 응용한 "방법 1"은 모형을 가정한 엔트로피에 기초한 최적 기준으로 검토한 경우 우수하였다. "방법 2"는 표본조사에 널리 쓰이는 계통추출을 응용하였으며 입력변수가 각기 다른 9개의 실험함수에 관하여 표본 평균의 추정치와 분산, MSE를 비교한 결과, 다른 실험계획들보다 우수하였다. 이 결과는 실험점이 실험영역 전체에서 골고루 퍼져서 나타난 것으로 보이며, 향후 전산실험계획에서의 응용을 기대할 수 있다.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.44 no.1
• /
• pp.26-32
• /
• 2007

In sequential speaker classification, the lack of prior information about the speakers poses a challenge for model initialization. To address the challenge, a predetermined generic model set, called Sample Speaker Models, was previously proposed. This approach can be useful for accurate speaker modeling without requiring initial speaker data. However, an optimal method for sampling the models from a generic model pool is still required. To solve this problem, the Speaker Quantization method, motivated by vector quantization, is proposed. Experimental results showed that the new approach outperformed the random sampling approach with 25% relative improvement in error rate on switchboard telephone conversations.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.30 no.4B
• /
• pp.367-377
• /
• 2010

In this study, we considered annual maximum rainfall data from 56 weather stations for rainfall frequency analysis using SIR(Sampling Important Resampling) algorithm and Bootstrap method. SIR algorithm is resampling method considering weight in extreme rainfall sample and Bootstrap method is resampling method without considering weight in rainfall sample. Therefore we can consider the difference between SIR and Bootstrap method may be due to the climate change. After the frequency analysis, we compared the results. Then we derived the results which the frequency based rainfall obtained using the data from SIR algorithm has the values of -10%~60% of the rainfall obtained using the data from Bootstrap method.

• Journal of the Korean Data and Information Science Society
• /
• v.26 no.3
• /
• pp.593-610
• /
• 2015

Adaptive Rejection Sampling (ARS) method is a well-known random number generator to acquire a random sample from a probability distribution, and has the advantage of improving the proposal distribution during the sampling procedures, which update it closer to the target distribution. However, the use of ARS is limited since it can be used only for the target distribution in the form of the log-concave function, and thus various methods have been proposed to overcome such a limitation of ARS. In this paper, we attempt to compare five random number generators based on ARS in terms of adequacy and efficiency. Based on empirical analysis using simulations, we discuss their results and make a comparison of five ARS-based methods.

• The Korean Journal of Applied Statistics
• /
• v.30 no.6
• /
• pp.993-1004
• /
• 2017

A stratified sampling method is generally used with a sample selected using the same sample weight in each stratum in order to improve the accuracy of the sampling survey estimation. However, the weight should be adjusted to reflect the response rate if the response rate is affected by the value of the variable of interest. It may be also more effective to adjust the weights by subdividing the stratum rather than using the same weight if the variable of interest has a linear relationship with the continuous auxiliary variables. In this study, we propose a method to increase the accuracy of estimation using an informative sampling design technique when the response rate is an exponential function of the variable of interest and the variable of interest has a linear relationship with the auxiliary variable. Simulation results show the superiority of the proposed method.

• Korean journal of applied entomology
• /
• v.20 no.4 s.49
• /
• pp.217-222
• /
• 1981

As a primary study on the economic injury level of A. tokionis larval populations in tobacco fields, we carried out some experiments for the evaluation of sampling efficiency of 3 sampling methods (pit-fall traps of apple pomace and rice bran, and clover patches) for the larval population and the regressions between loss and infested larval density The $10\times10cm$ clover patch showed a better sampling efficiency $11.8\;to\;18.0\%$ than the others. The sampling efficiency of clover patch becomes higher when the plots did not have any green plants. The linear regression equation (Y=4.2+1.383x) between loss (Y:kg/10a) and infested larval density (X: no. of larvae/plot) which was obtained by substitution of damage ratio and corrected damage ratio fitted the observed data better than the one (Y=2.68X) obtained without substitioning.

• Proceedings of the Korean Association for Survey Research Conference
• /
• 2006.12a
• /
• pp.11-40
• /
• 2006

시간의 흐름에 따라 사회 구성원들에 대한 행태 연구나 사회의 변화가 개인의 행동양식에 미치는 영향 등에 대한 조사에서는 어느 한 시점에서의 구성원들에 대한 횡단면 조사와는 다르게 다년간 지속적으로 조사개체를 추적조사 해야 하는 종단면 조사 또는 패널조사를 수행해야 한다. 패널조사는 횡단면 조사와는 달리 최초 표본이 시간이 지남에 따라 조사 대상 표본으로부터 탈락함으로서 발생하는 표본의 마모와 그에 따른 대표성 상실의 문제이다. 그러므로 이러한 표본의 대표성 상실 문제를 적절히 해결하기 위해 적용 가능한 방법이 가중치 조정 방법이다 횡단면 조사에서는 (1)추출가중치의 조정, (2)무응답 가중치 조정, (3)사후층화 가중치 조정과 같이 3단계의 가중치 조정과정을 수행하지만, 패널 조사의 경우 이와 더불어 원 표본의 대표성을 유지하기 위해 종단면 가중치(longitudinal weight)를 함께 고려해야 한다. 이러한 관점에서 본 연구에서는 다양한 패널형태에 따른 가중치 조정 방법에 대해 고찰하고, 향후 수행될 한국복지패널(Korea Welfare Panel Study: KWPS)의 가중치 산정에 관한 이론적 근거를 마련함과 동시에 현재 국내에서 수행되고 있는 패널조사의 가중치 조정방법과 비교하고자 한다.

• The Korean Journal of Applied Statistics
• /
• v.28 no.2
• /
• pp.175-188
• /
• 2015

This study is about estimation methods for the population pharmacokinetic and pharmacodymic model. This is a nonlinear mixed effect model, and it is difficult to find estimates of parameters because of nonlinearity. In this study, we examined theoretical background of various estimation methods provided by NONMEM, which is the most widely used software in the pharmacometrics area. We focused on estimation methods using a stochastic sampling approach - IMP, IMPMAP, SAEM and BAYES. The SAEM method showed the best performance among methods, and IMPMAP and BAYES methods showed slightly less performance than SAEM. The major obstacle to a stochastic sampling approach is the running time to find solution. We propose new approach to find more precise initial values using an ITS method to shorten the running time.

• The Korean Journal of Applied Statistics
• /
• v.27 no.1
• /
• pp.43-59
• /
• 2014

Cut-off sampling has been widely used for a highly skewed population like a business survey by discarding a part of the population, so called a take-nothing stratum. For a more accurate estimate of the population total, Hwang and Shin (2013) suggested a composite estimator of a take-nothing stratum total that combined the survey results of a take-nothing stratum and a take-some sub-stratum (a part of take-some stratum). In this paper we propose a new cut-off sampling scheme by considering a cost function and a composite estimator based on the proposed sampling scheme. Small simulation studies compared the performances of known composite estimators and the new composite estimator suggested in this study. We also use Briquette Consumption Survey data for real data analysis.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.5
• /
• pp.2251-2258
• /
• 2012

This study compared the unweighted sample design, frequency weighted sample design and complex sample design to using 2009 Korea National Health and Nutrition Examination Survey in an effort to identify whether or not there is any difference in potential risk factors. Pearson chi-square test and Rao-scott chi-square test were applied to the analytic methods. As a result of analyses, all the variables were overestimated as significant risk factors in case of the unweighted sample design to which only the frequency weights were applied. In addition, there were differences in the confidence levels and results from the simple random sampling analysis and complex sample design to which no weight was applied. It is necessary to carry out the complex sample design rather than the analysis to which the frequency weights are applied, in order to ensure the findings to represent the whole population when our national statistics data is used.