• The Korean Journal of Applied Statistics
• /
• v.30 no.2
• /
• pp.291-299
• /
• 2017

Mack and Skillings (1980) proposed nonparametric procedures in a randomized block design with replications as general alternatives. This method is used to find the difference in the treatment effect; however, it can cause a loss of inter block information using the ranking in each block. In this paper, we proposed new nonparametric procedures in a randomized block design with replications using an aligned method proposed by Hodges and Lehmann (1962) that used information of blocks and based on the joint placement suggest by Chung and Kim (2008). We also compared the power of the test of the proposed procedures and established a method through Monte Carlo simulation.

• Journal of the Korean Data and Information Science Society
• /
• v.26 no.3
• /
• pp.549-559
• /
• 2015

The MDPDE is an attractive alternative to maximum likelihood estimator because of the strong robustness properties that it inherently possess. The characteristics of MDPDE can be varied with the tuning parameter, in general, there is a trade-off between robustness and asymptotic efficiency. Hence, selection of optimal tuning parameter is important but complicated task. In this study, we introduce two optimal tuning parameter selection methods proposed by Fujisawa and Eguchi (2005) and Warwick (2006). Through simulation study, we found out that Warwick's method yields excessively small optimal tuning parameter in certain cases while Fujisawa and Eguchi's method performs well. Therefore, we think Fujisawa and Eguchi's method can be used commonly for finding optimal tuning parameter of MDPDE.

• Communications for Statistical Applications and Methods
• /
• v.15 no.2
• /
• pp.283-292
• /
• 2008

Smoothing parameter or bandwidth plays a key role in nonparametric classification based on kernel density estimation. We consider choosing smoothing parameter in nonparametric classification, which optimize the Bayes risk. Hall and Kang (2005) clarified the theoretical properties of smoothing parameter in terms of minimizing Bayes risk and derived the optimal order of it. Bootstrap method was used in their exploring numerical properties. We compare cross-validation and bootstrap method numerically in terms of optimal order of bandwidth. Effects on misclassification rate are also examined. We confirm that bootstrap method is superior to cross-validation in both cases.

• Environmental and Resource Economics Review
• /
• v.25 no.1
• /
• pp.1-25
• /
• 2016

This study investigated the variability of WTP estimates(i.e. mean or median) with ad hoc assumptions of specific parametric probability distributions(i.e. normal, logistic, lognormal, and exponential distribution) to estimate WTP function using dichotomous choice CV data on mortality risk reduction. From the perspective of policy decision, the variability of these WTP estimates are intolerable in comparison with those of Turnbull nonparametric estimation method which is free from ad hoc distribution assumptions. The Turnbull nonparametric estimation can avoid a kind of misspecification bias due to ad hoc assumption of specific parametric distributions. Furthermore, the WTP estimates by Turnbull nonparametric estimation are robust because the similar estimates are elicited from a dichotomous choice or double dichotomous choice CV data, and the statistically significant WTP estimates can be obtained even though it is not possible by parametric estimation methods. If there are considerable variability among those WTP estimates by parametric estimation methods in condition with no criteria of model adequacy, the mean WTPs from Turnbull nonparametric estimation can be the robust estimates without ad hoc assumptions, which can avoid controversial issues in the perspective of policy decisions.

• The Korean Journal of Applied Statistics
• /
• v.20 no.2
• /
• pp.357-371
• /
• 2007

For the time-to-failure data with competing risks, cumulative incidence functions (CIFs) are commonly estimated using nonparametric methods. If the cases of events due to the cause of primary interest are infrequent relative to other cause of failure, nonparametric methods may result in rather imprecise estimates for CIF. In such cases, Bryant et al. (2004) suggested to model the cause-specific hazard of primary interest parametrically, while accounting for the other modes of failure using nonparametric estimator. We represented the semiparametric cumulative incidence estimator and extended to the model of Weibull and log-normal distribution. We also conducted simulations to access the performance of the semiparametric cumulative incidence estimators and to investigate the impact of model misspecification in log-normal cause-specific hazard model.

• Communications for Statistical Applications and Methods
• /
• v.5 no.1
• /
• pp.121-131
• /
• 1998

이 논문의 목적은 비모수 회귀모형에 있어서의 오차의 분산을 추정하는 방법들 중 차분에 기저한 방법 (difference-based methods)을 이용한 기존의 추정량들을 비교 분석하는데 있다. 특히 점근적인 최적 이차 차분에 기저한 Hall과 Kay, Titterington(1990)의 HKT 추정량에 대한 그들의 추정량에 대한 문제점들을 제시하고, HKT추정량과, GSJS추정량, Rice추정량에 대하여 모의 실험을 이용하여 모수에 대한 수렴 속도를 비교 분석 하였다. 또한 GSJS 추정량에 대한 일치성과 수렴 속도를 보였다.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.30 no.1
• /
• pp.87-94
• /
• 2017

Reliability analysis(RA) and Reliability-based design optimization(RBDO) require statistical modeling of input random variables, which is parametrically or nonparametrically determined based on experimental data. For the parametric method, goodness-of-fit (GOF) test and model selection method are widely used, and a sequential statistical modeling method combining the merits of the two methods has been recently proposed. Kernel density estimation(KDE) is often used as a nonparametric method, and it well describes a distribution function when the number of data is small or a density function has multimodal distribution. Although accurate statistical models are needed to obtain accurate RA and RBDO results, accurate statistical modeling is difficult when the number of data is small. In this study, the accuracy of two statistical modeling methods, SSM and KDE, were compared according to the number of data. Through numerical examples, the RA results using the input models modeled by two methods were compared, and appropriate modeling method was proposed according to the number of data.

• Communications for Statistical Applications and Methods
• /
• v.16 no.3
• /
• pp.463-477
• /
• 2009

Parameter estimation methods such as maximum likelihood estimation method, probability weighted moments method, regression method have been popularly applied to various extreme value models in numerous literature. Among three methods above, the performance of regression method has not been rigorously investigated yet. In this paper the regression method is compared with the other methods via Monte Carlo simulation studies for estimation of parameters of the Generalized Extreme Value(GEV) distribution and the Generalized Pareto(GP) distribution. Our simulation results indicate that the regression method tends to outperform other methods under small samples by providing smaller biases and root mean square errors for estimation of location parameter of the GEV model. For the scale parameter estimation of the GP model under small samples, the regression method tends to report smaller biases than the other methods. The regression method tends to be superior to other methods for the shape parameter estimation of the GEV model and GP model when the shape parameter is -0.4 under small and moderately large samples.

• The Korean Journal of Applied Statistics
• /
• v.27 no.4
• /
• pp.553-560
• /
• 2014

This paper deals with estimable functions of parameters of less than full rank linear model. In general, the parameters of an overspecified model are not uniquely determined by least squares solutions. It discusses how to formulate linear estimable functions as functions of parameters in the model and shows how to use projection matrices to check out whether a parameter or function of the pamameters is estimable. It also presents a method to form a basis set of estimable functions using linearly independent characteristic vectors generating the row space of the model matrix.

• The Korean Journal of Applied Statistics
• /
• v.26 no.5
• /
• pp.713-723
• /
• 2013

An objective Bayesian estimation procedure of the two-parameter Pareto distribution is presented under the reference prior and the noninformative prior. Bayesian estimators are obtained by Gibbs sampling. The steps to generate parameters in the Gibbs sampler are from the shape parameter of the gamma distribution and then the scale parameter by the adaptive rejection sampling algorism. A numerical study shows that the proposed objective Bayesian estimation outperforms other estimations in simulated bias and mean squared error.