• 한국데이터정보과학회:학술대회논문집
• /
• 한국데이터정보과학회 2001년도 추계학술대회
• /
• pp.67-75
• /
• 2001

A completely randomized growth curve model was defined by Zerbe(1979). We propose the fully significant difference procedure for multiple comparisons of completely randomized growth curve model. The standard F test is useful tool to multiple comparisons of the completely randomized growth curve model. The proposed method is applied to experimental data.

• Journal of the Korean Data and Information Science Society
• /
• 제14권2호
• /
• pp.313-323
• /
• 2003

For the growth curve model with arbitrary covariance structure, known as unstructured covariance matrix, the problems of detecting outliers are discussed in this paper. In order to detect outliers in the growth curve model, the test statistics using U-distribution is established. After detecting outliers in growth curve model, we test homo and/or hetero-geneous covariance matrices using PSR Quasi-Bayes Criterion. For illustration, one numerical example is discussed, which compares between before and after outlier deleting.

• 정보처리학회논문지D
• /
• 제11D권3호
• /
• pp.597-604
• /
• 2004

정확한 소프트웨어 비용 추정은 개발자와 고객 모두에게 중요하다. 대부분의 비움 추정 모델들은 규모 추정으로부터 틴은 라인 수와 기능점수와 같을 규모 측도에 기반을 두고 있다. 규모 추정의 정확도는 비용 추정 정확도에 직접적으로 영향을 미친다. 이에 따라 대부분의 회귀기반 비용추정 모델들은 규모에 기반한 멱함수 형태를 적용하고 있다. 생물의 성장, 기술의 발전과 인간의 학습 능력 등 많은 성장 현상들은 S자 곡선을 따른다. 본 논문은 성장곡선을 이용하여 개발노력을 추정하는 모델을 제시하였다. 제시된 모델은 소프트웨어 규모가 증가함에 따라 소요되는 개발 비용이 성장곡선을 따른다고 가정한다. 일반적인 소프트웨어 규모 추정 기법인 기능점수, 완전기능점수와 유스케이스 점수에 기반하여 성장곡선 모델의 적합성을 검증하였다. 제안된 성장곡선 모델들은 멱함수 모델과 비교 시 상호 견줄만한 성능을 보여 소프트웨어 비용 추정분야에 석용 가능함을 보였다.

• 산업공학
• /
• 제23권3호
• /
• pp.205-212
• /
• 2010

The learning curve model is a mathematical form which represents the relationship between the manufacturing experience and its effectiveness. The semiconductor manufacturing is widely known as an appropriate example for the learning effect due to its complicated manufacturing processes. In this paper, I propose a new compound learning curve model for semiconductor products in which the general learning curve model and the growth curve are composed. The dependent variable and the effective independent variables of the model were abstracted from the existing learning curve models and selected according to multiple regression processes. The simulation results using the historical DRAM data show that the proposed compound learning curve model is one of adequate models for describing learning effect of semiconductor products.

• Journal of the Korean Data and Information Science Society
• /
• 제10권2호
• /
• pp.369-385
• /
• 1999

성장곡선모형에서 다중 이상값들이나 영향관측값들을 탐지하는 문제는 선형회귀모형에서의 문제에 비해 매우 복잡하여 거의 이루어지지 않고 있는 실정이다. 본 연구에서는 이상점을 포함하고 있는 성장곡선모형에서 이들을 탐지하는 방법으로 평균이동모형을 이용하는 방법을 소개하였다. 이 방법을 이용하여 찾아낸 자료가 이상점인지의 여부를 예측표본재이용 의사 베이즈 우도 기준법을 이용한 등분산성의 검정을 통해 알아보았다. 끝으로 Potthoff(1964)등이 사용한 자료를 이용한 예제를 통해 이상점 탐지와 등분 산성 검정을 실시한 결과를 제시하였다.

• 농업과학연구
• /
• 제43권4호
• /
• pp.623-633
• /
• 2016

A total of 6,973 steer growth records of Hanwoo breeding bull's progeny test data collected from 1989 to 2015 were analyzed to identify the most appropriate growth curve among three growth curve models (Gompertz, Logistic and von Bertalanffy). The Gompertz growth curve model equation was $W_t=990.5e^{{-2.7479e}^{-0.00241t}}$, the Logistic growth curve model equation was $W_t=772(1+8.3314e^{-0.00475t})^{-1}$, and the von Bertalanffy growth curve model equation was $W_t=1,196.4(1-0.646e^{-0.00162t})^3$. The Gompertz model parameters A, b, and k were estimated to be $990.5{\pm}10.27$, $2.7479{\pm}0.0068$, and $0.00241{\pm}0.000028$, respectively. The inflection point age was estimated to be 421 days and the weight of inflection point was 365.3 kg. The Logistic model parameters A, b, and k were estimated to be $772.0{\pm}4.12$, $8.3314{\pm}0.0453$, and $0.00475{\pm}0.000033$, respectively. The inflection point age was estimated to be 445 days and the weight of inflection point was 385.0 kg. The von Bertalanffy model parameters A, b, and k were estimated to be $1196.4{\pm}18.39$, $0.646{\pm}0.0010$, and $0.00162{\pm}0.000027$, respectively. The inflection point age was estimated to be 405 days and the weight of inflection point was 352.0 kg. Mature body weight of the von Bertalanffy model was 1196.4 kg, the Gompertz model was 990.5 kg, and the Logistic model was 772.0 kg. The difference between actual and estimated weights was similar in the Logistic model and the von Bertalanffy model. The difference between market weight and estimated market weight was the lowest in the Gompertz model. The growth curve using the von Bertalanffy model showed the lowest mean square error.

• 정보처리학회논문지D
• /
• 제11D권7호
• /
• pp.1451-1458
• /
• 2004

Gompertz 성장곡선에 기반한 기존의 소프트웨어 신뢰성 성장모델들은 모두 대수형이다. 대수형 Gompertz 성장 곡선에 기반한 소프트웨어 신뢰성 성장 모델들은 모수 추정에 어려움을 갖고 있다. 그러므로 본 논문은 로지스틱형 Gompertz 성장곡선에 기반한 신뢰성 성장 모델을 제안한다. 13개의 다른 소프트웨어 프로젝트로부터 얻은 고장 데이터를 분석하여 그 유용성을 검토하였다. 모델의 모수들은 변수변환을 통한 선형희귀분석과 Virence의 방법으로 추정되었다. 제안된 모델은 평균 상대 예측 오차에 기반하여 성능을 비교하였다. 실험 결과 제안된 모델은 대수형 Gompertz 성장 곡선에 기반한 모델보다 좋은 성능을 보였다.

• 품질경영학회지
• /
• 제23권1호
• /
• pp.95-105
• /
• 1995

The error structure of nonlinearized technological growth models, such as, the Pearl curve, the Gompertz curve and the Wei bull growth curve, has zero mean and a constant variance over time. Transformed models, however, like the linearized Fisher-Pry model. the linearized Gompertz growth curve, and the linearized Weibull growth curve have increasing variance from t = 0 to the inflection point.

• 한국환경과학회지
• /
• 제31권9호
• /
• pp.793-801
• /
• 2022

This study was carried out to provide basic data for logical forest management by developing a site index curve reflecting the current growth characteristics of Cryptomeria japonica stands in Korea. The height growth model was developed using the Chapman-Richards, Schumacher, Gompertz, and Weibull algebraic difference equations, which are widely used in growth estimation, for data collected from 119 plots through the 7th National Forest Inventory and stand survey. The Chapman-Richards equation, with the highest model fit, was selected as the best equation for the height growth model, and a site index curve was developed using the guide curve method. To compare the developed site index curve with that on the yield table, paired T-tests with a significance level of 5% were performed. The results indicated that there were no significant differences between the site index curve values at all ages, and the p-value was smaller after the reference age than before. Therefore, the site index curve developed through this study reflects the characteristics of the changing growth environment of C. japonica stands and can be used in accordance with the site index curve on the current yield table. Thus, this information can be considered valuable as basic data for reasonable forest management.

• 품질경영학회지
• /
• 제25권1호
• /
• pp.44-55
• /
• 1997

In order to determine procedures for a, pp.opriate model selection of technological growth curves, numerous time series that were representative of growth behavior were collected according to data characteristics. Three different growth curve models were fitted onto data sets in an attempt to determine which growth curve models achieved the best forecasts for types of growth data. The analysis of the results gives rise to an a, pp.oach for selecting a, pp.opriate growth curve models for a given set of data, prior to fitting the models, based on the characteristics of the goodness of fit test.