• The Korean Journal of Applied Statistics
• /
• v.21 no.3
• /
• pp.441-453
• /
• 2008

We analyzed Korean professional basketball and baseball players salary under the assumption that it depends on the personal records and contribution to the team in the previous year. We extensively used data visualization tools to check the relationship among the variables, to find outliers and to do model diagnostics. We used multiple linear regression and regression tree to fit the model and used cross-validation to find an optimal model. We check the relationship between variables carefully and chose a set of variables for the stepwise regression instead of using all variables. We found that points per game, number of assists, number of free throw successes, career are important variables for the basketball players. For the baseball pitchers, career, number of strike-outs per 9 innings, ERA, number of homeruns are important variables. For the baseball hitters, career, number of hits, FA are important variables.

• The Korean Journal of Applied Statistics
• /
• v.34 no.2
• /
• pp.153-166
• /
• 2021

A main goal of pharmacogenomics studies is to predict individual's drug responsiveness based on high dimensional genetic variables. Due to a large number of variables, feature selection is required in order to reduce the number of variables. The selected features are used to construct a predictive model using machine learning algorithms. In the present study, we applied several hybrid feature selection methods such as combinations of logistic regression, ReliefF, TurF, random forest, and LASSO to a next generation sequencing data set of 400 epilepsy patients. We then applied the selected features to machine learning methods including random forest, gradient boosting, and support vector machine as well as a stacking ensemble method. Our results showed that the stacking model with a hybrid feature selection of random forest and ReliefF performs better than with other combinations of approaches. Based on a 5-fold cross validation partition, the mean test accuracy value of the best model was 0.727 and the mean test AUC value of the best model was 0.761. It also appeared that the stacking models outperform than single machine learning predictive models when using the same selected features.

• Journal of Life Science
• /
• v.27 no.1
• /
• pp.38-43
• /
• 2017

Glycosyl aesculin, a potent anti-inflammatory agent, was synthesized by transglycosylation reaction, catalyzed by Thermotoga neapolitana ${\beta}-glucosidase$, with aesculin as an acceptor. The key reaction parameters were optimized using response-surface methodology (RSM) and $2{\mu}g$ of the enzyme. As shown by a statistical analysis, a second-order polynomial model fitted well to the data (p<0.05). The response surface curve for the interaction between aesculin and other parameters revealed that the aesculin concentration and reaction time were the primary factors that affected the yield of glycosyl aesculin. Among the tested factors, the optimum values for glycosyl aesculin production were as follows: aesculin concentration of 9.5 g/l, temperature of $84^{\circ}C$, reaction time of 81 min, and pH of 8.2. Under these conditions, 61.7% of glycosyl aesculin was obtained, with a predicted yield of 5.86 g/l. The maximum amount of glycosyl aesculin was 6.02 g/l. Good agreement between the predicted and experimental results confirmed the validity of the RSM. The optimization of reaction conditions by RSM resulted in a 1.6-fold increase in the production of glycosyl aesculin as compared to the yield before optimization. These results indicate that RSM can be effectively used for process optimization in the synthesis of a variety of biologically active glycosides using bacterial glycosidases.

• Korean Chemical Engineering Research
• /
• v.55 no.5
• /
• pp.723-730
• /
• 2017

The batch experiments by response surface methodology (RSM) have been applied to investigate the influences of operating parameters such as temperature, initial concentration, contact time and adsorbent dosage on 2,4-dichlorophenol (2,4-DCP) adsorption with an activated carbon prepared from waste citrus peel (WCAC). Regression equation formulated for the 2,4-DCP adsorption was represented as a function of response variables. Adequacy of the model was tested by the correlation between experimental and predicted values of the response. A fairly high value of $R^2$ (0.9921) indicated that most of the data variation was explained by the regression model. The significance of independent variables and their interactions were tested by the analysis of variance (ANOVA) and t-test statistics. These results showed that the model used to fit response variables was significant and adequate to represent the relationship between the response and the independent variables. The kinetics and isotherm experiment data can be well described with the pseudo-second order model and the Langmuir isotherm model, respectively. The maximum adsorption capacity of 2,4-DCP on WCAC calculated from the Langmuir isotherm model was 345.49 mg/g. The rate controlling mechanism study revealed that film diffusion and intraparticle diffusion were simultaneously occurring during the adsorption process. The thermodynamic parameters indicated that the adsorption reaction of 2,4-DCP on WCAC was an endothermic and spontaneous process.

• Food Science and Preservation
• /
• v.13 no.3
• /
• pp.363-368
• /
• 2006

This study was performed to develop flavor materials from Elsholtzia ciliata Hylander with analyzing functionality and aroma profile and to optimize supercritical fluid extraction method and optimum condition. The qualities of water extracts such as total yield total phenolic compound electron donation ability, estragole and L-carvone, were affected by extraction pressure than time. The response variables had significant with pressure than with time and the established polynomial model was suitable(P>0.05) model by Lack-of-Fit analysis. The optimum extraction conditions which were limited of maximum value for dependent variables under experimental conditions based on central composite design were 238 bar and 42 min.

• The Korean Journal of Applied Statistics
• /
• v.32 no.2
• /
• pp.291-300
• /
• 2019

This study investigates the characteristics of evaluation measures for classification models on a binary response variable in order to evaluate their suitability for use. Six measures are considered: Accuracy, Sensitivity, Specificity, Precision, F-measure, and the Heidke's skill score (HSS). Evaluation measures are reformulated using x(ratio of actually 1), y(ratio predicted by 1), z(ratio of both actual and predicted by 1) from the confusion matrix. We suggest two necessary conditions to assess the suitability of the evaluation measures. The first condition is that the measure function is constant for x and y in the case of a random model. The second condition is that the measure function is increasing for z and decreasing for x and y. Since only HSS satisfies the two conditions, that is always appropriate as an evaluation measure for the classification model on the binary response variable, and the other measures should be used within a limited range.

• Industry Promotion Research
• /
• v.6 no.1
• /
• pp.1-7
• /
• 2021

We investigated multi-layer perceptrons and supervised learning algorithms, and also examined how to model functional relationships between covariates and response variables using a package called neuralnet. The algorithm applied in this paper is characterized by continuous adjustment of the weights, which are parameters to minimize the error function based on the comparison between the actual and predicted values of the response variable. In the neuralnet package, the activation and error functions can be appropriately selected according to the given situation, and the remaining parameters can be set as default values. As a result of using the neuralnet package for the infertility data, we found that age has little influence on infertility among the four independent variables. In addition, the weight of the neural network takes various values from -751.6 to 7.25, and the intercepts of the first hidden layer are -92.6 and 7.25, and the weights for the covariates age, parity, induced, and spontaneous to the first hidden neuron are identified as 3.17, -5.20, -36.82, and -751.6.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2020.06a
• /
• pp.148-148
• /
• 2020

산업의 고도화가 진행됨에 따라 화학원료의 사용이 증가하고 있고 독성을 가진 화학물질이 하천으로 유입되는 사고가 빈번하게 발생하고 있다. 수환경으로 유입되는 유해화학물질은 주로 무색무취의 물질들로 사고가 발생하더라도 초기 발견이 어려워 어류폐사를 유발하거나 취수시설에서 용수로 취수되는 경우가 발생하기 때문에 이에 대한 대응책 마련이 필수적이다. 하천에 유입된 오염물질의 거동을 신속하게 예측하기 위해 1차원 오염물질 추적 모형이 활용되는데, Fickian 이송-분산 모형(Fickian Advection-dispersion equation model; FADE)이 주로 사용되고 있다. 하지만 FADE는 오염물질이 하천 저장대에서 지체되는 현상을 반영하지 못하기 때문에 농도곡선의 왜곡도를 구현하지 못하는 단점을 가지고 있다. 따라서 본 연구에서는 하천저장대모형(River Storage Model; RSM)을 개발하고 이를 국가하천인 감천에 적용하였다. 본 연구에서 개발한 RSM은 분산계수, 본류대 면적, 저장대 면적, 저장대 교환계수의 네 가지 매개변수를 통해서 하천의 물질 저장 및 교환 특성를 구현한 non-Fickian 모형으로서, 생화학반응, 휘발, 흡·탈착항을 추가하여 유해화학물질의 혼합 거동을 정확하게 모의할 수 있도록 개발하였다. 저장대 모형의 매개변수를 산정하기 위해서 하천 유량과 지형자료를 기반으로 HEC-RAS를 모의하여 계산된 수리특성을 입력변수로 사용하였다. 저수기, 평수기, 풍수기 유량을 기준으로 세 경우의 시나리오 모의를 수행하였는데, 5ton의 톨루엔이 김천산업단지에서 감천으로 유입된 경우 약 20km 하류에 위치한 취수장에서 톨루엔의 농도변화를 예측했다. 보존성 물질에 대한 모의 결과, 풍수기의 경우 저수기에 비해 유속이 크기 때문에 취수장에서 20.56시간 먼저 기준농도에 도달하고, 7.21시간 더 짧게 머무는 것으로 나타났다. 유해화학물질의 반응특성에 대한 민감도 분석을 시행한 결과, 생화학적 반감기가 18.98시간보다 길고, 옥탄올-물 분배계수가 2.267 이하인 물질은 생분해 및 흡·탈착 반응에 둔감한 것으로 나타났다. 1m 수심 기준 0.114m/s 이하 유속에서의 하천 수리조건에서는 화학물질의 휘발성을 무시할 수 있는 것으로 밝혀졌다.

• Journal of the Korean Society of Propulsion Engineers
• /
• v.22 no.3
• /
• pp.21-27
• /
• 2018

In this paper the laminate design parameters are researched to maximize the performance index of a composite pressure vessel. To maximize the performance index, the three design variables that the thickness of each of helical and hoop layers and the length of hoop layer are considered under the assumption of fixed internal space. To optimize the variables, the response surface method is introduced for construction of the surrogate model and the ANOVA(analysis of variance) is performed to evaluate the effects of the variables. The optimization problem is formulated to maximize performance index under the burst pressure constraint. To verify the effectiveness of the research, numerical analyses are performed for the optimum model.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2017.05a
• /
• pp.24-24
• /
• 2017

홍수와 가뭄 등 수문기상재해 분석 및 사전 예측하기 위해서는 강수뿐만 아니라 토양수분, 증발산, 유량, 등과 같이 지표?하의 수문기상정보를 고려하는 것이 필요하다. 본 연구에서는 National Center for Atmospheric Research (NCAR)에서 개발된 고해상도 수문기상정보 모의가 가능한 WRF-Hydro를 활용하여 남강댐 유역에서 발생되는 돌발홍수 예측 적용성 평가를 수행하였다. 모델의 시공간 해상도는 1 hr, 150 m 이며, 기상 관측자료(Automatic Weather System, Automated Synoptic Observing System)를 사용하여 매개변수 민감도 실험을 실시하여 최적 모델 설정을 제시하였다. 고려된 매개변수는 격자 침투량을 결정하는 변수, 초기 저류 깊이, 표면 저항계수, 조도계수와 초기 토양수분 정보이며, 검증에 사용된 정보는 국가수자원관리종합정보시스템에서 1시간 단위로 제공되는 유입량 정보를 사용하였다. 그 결과 유출량은 격자 침투량을 결정하는 변수와 조도계수에 따라 민감하게 반응하였으며, 초기 토양수분량의 변화에 따라 시간에 따른 유출량의 변화가 강수에 민감하게 반응하는 것을 확인 할 수 있었다. 보정된 매개변수를 적용하여 돌발홍수 신고 지점의 유출량 변화를 살펴본 결과 강수의 발생과 동시에 매우 빠르게 유출량이 발생한 것을 볼 수 있었다.