• Journal of the Korean Statistical Society
• /
• 제17권2호
• /
• pp.61-80
• /
• 1988

For testing homogeneity of scale parameters aginst ordered alternatives, some nonparametric test statistics based on pairwise ranking method are proposed. The proposed tests are distribution-free. The asymptotic distributions of the proposed statistcs are also investigated. It is shown that the Pitman efficiencies of the proposed rank tests are the same as those of the corresponding two-sample rank tests in the scale problem. A small-sample Monte Carlo study is also performed. The results show that the proposed tests are robust and efficient.

• 응용통계연구
• /
• 제12권2호
• /
• pp.505-515
• /
• 1999

인공신경망 모형을 적합시키는데 사용하는 역전파 알고리즘을 로버스트하게 만드는 새로운 오차함수를 제안했으며, 새 방법의 성능을 확인하기 위해 Liano가 제안한 방법에 따라 모의실험을 수행했다. 실험결과 새 방법은 LMS방법만큼 안정적이었으며, Liano의 LMLS방법보다 더 로버스트했다. 또 실제 사례를 분석함으로써 이 방법이 의미있는 방법임을 보였다. 새 방법은 특히 오차가 없거나 작은 오차를 갖는 표본에 대해서도 좋은 성질을 가짐으로서 대형오차의 유무에 관계없이 항상 사용할 수 있는 방법으로 판명되었다.

• 응용통계연구
• /
• 제12권2호
• /
• pp.479-490
• /
• 1999

선형시계열모형인 AR(1)모형과 비선형시계열모형인 RCA(1), ARCH(1)모형에서 이상치(Outlier)가 존재할 경우 최소제곱추정량과 M추정량간의 점근상대효율(Asymptotic Relative Efficiency: ARE)을 구하여 두 추정량의 로버스트 성질을 비교·분석하였다. 또한 여러 유계함수(Huber, Tukey, Andrews, Hampel)들을 M추정함수에 적용하여 각각의 유계함수들을 비교·분석하였다.

• Communications for Statistical Applications and Methods
• /
• 제12권3호
• /
• pp.659-672
• /
• 2005

This paper evaluates discretization of continuous variables to select relevant variables for supervised learning using mutual information. Three discretization methods, MDL, Histogram and 4-Intervals are considered. The process of discretization and variable subset selection is evaluated according to the classification accuracies with the 6 real data sets of UCI databases. Results show that 4-Interval discretization method based on quantiles, is robust and efficient for variable selection process. We also visually evaluate the appropriateness of the selected subset of variables.

• Communications for Statistical Applications and Methods
• /
• 제15권1호
• /
• pp.95-104
• /
• 2008

This paper addresses issues of robustness in Bayesian optimal design. We may have difficulty applying Bayesian optimal design principles because of the uncertainty of prior distribution. When there are several plausible prior distributions and the efficiency of a design depends on the unknown prior distribution, robustness with respect to misspecification of prior distribution is required. We suggest a new optimal design criterion which has relatively high efficiencies across the class of plausible prior distributions. The criterion is applied to the problem of estimating the turning point of a quadratic regression, and both analytic and numerical results are shown to demonstrate its robustness.

• 응용통계연구
• /
• 제25권3호
• /
• pp.485-501
• /
• 2012

This research is to find a design D that minimizes forecast variance in d dimensions over the candidate space ${\chi}$. We want a robust design since we may not know the specific covariance structure. We seek a design that minimizes a coverage criterion and hope that this design will provide a small forecast variance even if the covariance structure is unobservable. The details of an exchange or swapping algorithm and several properties of the parameters of coverage criterion with the unknown correlation structures are discussed.

• Journal of the Korean Statistical Society
• /
• 제26권3호
• /
• pp.333-354
• /
• 1997

Computational algorithms to calculate M-estimators and rank estimators of regression parameters from left-truncated and right-censored data are developed herein. In the case of M-estimators, new statistical methods are also introduced to incorporate leverage assements and concomitant scale estimation in the presence of left truncation and right censoring on the observed response. Furthermore, graphical methods to examine the residuals from these data are presented. Two real data sets are used for illustration.

• 한국데이터정보과학회:학술대회논문집
• /
• 한국데이터정보과학회 2002년도 춘계학술대회
• /
• pp.97-106
• /
• 2002

Taguchi parameter design, the product-array approach using orthogonal arrays is mainly used. However, it often requires an excessive number of experiments. An alternative approach, which is called the combined-array approach, was suggested by Welch et. al. and studied by others. In these studies, only single quality characteristic was considered. We propose how to simultaneously optimize multiple quality characteristics using desirability function when we used the combined-array approach to assign control and noise factors. An example is illustrated to the combined-array approach.

• Communications for Statistical Applications and Methods
• /
• 제2권2호
• /
• pp.330-335
• /
• 1995

The rebustness of one sample Chi-square test for multivariate normal mean vector is investigated when the multivariate normal population is mixed with another multivariate normal population with differing in the mean vector. Explicit expressions for the level of significance and power of the test are derived. Some numerical results indicate that the Chi-square test procedure is quite robust against slight mixtures of multivariate normal populations differing in location parameters.

• Communications for Statistical Applications and Methods
• /
• 제2권2호
• /
• pp.13-19
• /
• 1995

This paper considers model robust accelerated life test plans for estimating the logmean or percentile of product lige which is exponentially distributed. A linear relationship between the log mean life and the stress is assumed as usual, while the true relationship is quadratic. Optimum plans are then obtained by minimizing asymptotic mean squared error of maximum likelihood estimator of the log mean life.