http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
성장곡선에서 교차타당성 법칙을 적용한 조정화 판별법에 관한 연구
심규박 동국대학교 자연과학연구원 1994 자연과학연구 논문집 Vol.14 No.-
The problem considered here is the optimal growth curve classification rule to be either singular or near singular. This paper deals with a way out of this problem. By using cross-validation method on regularized discriminant analysis, new classification for the growth curve model is suggested and examined via limited out informative simulation study.
심규박,류부형 동국대학교 경주대학 1996 東國論集 Vol.15 No.-
A commercial nuclear power station contains at least two emergency diesel generates (EDG) to control the risk of severe core damage during the station blackout accidents. Therefore the reliability of the EDG's to start and load-run on demand must be maintained at a sufficiently high level. Until now, a simple start and load-run success rates was used to calculate the EDG's reliability. However, this method has been many dafects. Recently, the work of Martz (1996) proposed a parametric emprical Bayes (PEB) estimator to find the EDG' s reliability. For binomial sampling model with pass/fail data and a beta prior distribution, PEB point and credibility interval estimates are provided for the EDG's reliability. In this paper, we will compare above two methods and show Martz's method is more reasonable in our situation with several examples.
심규박 동국대학교 대학원 1991 大學院硏究論集-東國大學校 大學院 Vol.21 No.-
A Sample-based procedure for selecting an optimal subset of variables for the multivariate dichotomous responses problem is presented. The procedure is based upon the nearest neighbor of order r=1 rule. For the selection criterion, we used the scaled average and scaled minimum method by N.G.lick. Some relative merits of this procedure are derived from a sampling experiment.
성장곡선모형의 균형판별분석에서 Poorly-posed 문제에 관한 연구
심규박,염준근 동국대학교 경주대학 1996 東國論集 Vol.15 No.-
Poorly-posed problems in the balanced discriminant analysis is considered. We restrict consideration to the case of observations and the number of variables are the same and small. When these problems exist, we do not use a maximum likelihood estimates(MLE) to estimate covariance matrices. Instead of MLE, an alternative estimate for the covariance matrices are proposed. This alternative method make good use of two regularization parameters. λ and γ. A new test rule for the discriminant function is suggested and examed via limited but informative simulation study. From the simulation study, it is shown that the suggested test rute gives better test result than other previously suggested method in terms of error rate criterion.
김혜중,심규박 동국대학교 자연과학연구소 1993 자연과학연구 논문집 Vol.13 No.-
소표본하의 성장곡선판별의 경우 최적 분류 법칙 추정시 발생되는 표본공분산 행렬의 비정칙문제가 판별의 정확성을 떨어뜨리는 것으로 지적되어 왔다. 본 논문은 이 문제의 해결책으로 표본공분산행렬의 조정화기법을 도입하여서 새로운 최적 성장곡선 분류법칙을 제안하였다. 제안된 분류법칙에 의한 판별법의 효율을 모의 실험에 의해 조사한 결과, 오분류오차 기준하에서 새로운 판별법이 기준의 것보다 우수할 뿐 아니라 비정칙 문제도 완전히 해결할 수 있는 방법임을 보였다. The problem considered here is the classification of the growth curve model useful especially for that with small training sample of sizes. It has long been pointed out that where we have small training samples, the samples, the sample convariance matrices for the optimal growth curve classification rule to be either singular or near singular. As a result, even there have been couple of well defined optimal classification rules, the problem of small training samples prevents the analyst to have correct classification result. This paper deals with a way out of this problem. By using a regulation technique of the sample covariance materices, new classification rule for the growth curve model is suggested study, it is shown that the suggested classification rule gives better classification than the other previously suggested method in terms of error rate criterion.
베이즈 예측분포법에 의한 線型回歸 模型의 異常點 탐지법 및 比較
김혜중,심규박 동국대학교 자연과학연구소 1990 자연과학연구 논문집 Vol.10 No.-
선형회귀 모형에서 이상점을 찾아내는 방법으로 베이지안 예측법을 제안하고자 한다. 이 방법을 사용하여 이상점을 구한 후, 기존 R.D. Cook(1977)의 방법과 K. cChaloner et al(1988)의 방법을 사용하여 구한 이상점을 비교하였다. 끝으로 예제를 통해 우리가 제시한 방법의 장점을 설명하였다. A bayesian predictive approach for detecting outliers in a linear model is developed. The suggested method is compared with the other existing methods ; R.D. Cook(1977)'s method and K. Chaloner et al.(1988)'s method. Some relative merits of our method are derived from the limited sampling experiment.
小標本 성장곡선모형에서 PSR 擬似 베이즈 기준의 적용
김혜중,심규박,이승주 동국대학교 자연과학연구소 1992 자연과학연구 논문집 Vol.12 No.-
WE suggest the Predictive Sample Reuse(PSR) Quasi-Bayes Criterion(QBC) in testing the equality form of covariance matrices and study the power of the test method on the small sampling Growth Curve Model. Based on the mean square error, We investigate the effect of QBC on estimating the Growth Curve. 소표본 성장곡선모형의 분석에서 필요한 공분산 행렬의 동일성 검정법으로 PSR ?似 베이즈 기준을 제시하였고, 이 검정법의 검정력에 대하여 연구하였다. 또한 제시된 검정 결과가 성장곡선모형의 추정에 어떤 영향을 미치는지를 평균제곱오차 기준으로 알아보았다.
1. 06㎛ Nd : YAG Laser를 이용한 요추간판탈출증의 경피적 수핵감압술
이흥선,변박장,최순관,신원한,권귀향,이인수,이경석,김범태 순천향대학교 1994 논문집 Vol.17 No.4
The 1.06㎛ Neodymium-YAG Laser(800 micron hemisherical) is revealed a useful tool for percutaneous laser disc decompression(PLDD). Especially PLDD would be applied to the management of lumbar disc herniation with minimal intervention. Total irradiated energy to one level of disc space was bout 1000 to 2000 joules. We performed this method at 10 levels, in 7 cases of patients, consisting of 4 males and 3 females with an average age of 37.7 years. Five cases showed excellent and good results, were observed. In two cases, open surgery was required because of unsatisfactory results. The best results were obtained in the protrusion type of disc herniation. We have summarized some advantages of PLDD: a) safe and simple procedure in an outpatient setting, b) avoiding general anesthesia, late spinal instability, c) possibility of repetitive use.