• Communications for Statistical Applications and Methods
• /
• v.24 no.4
• /
• pp.383-396
• /
• 2017

The model in our approach assumes that computer responses are a realization of a Gaussian processes superimposed on a regression model called a Gaussian process regression model (GPRM). Selecting a subset of variables or building a good reduced model in classical regression is an important process to identify variables influential to responses and for further analysis such as prediction or classification. One reason to select some variables in the prediction aspect is to prevent the over-fitting or under-fitting to data. The same reasoning and approach can be applicable to GPRM. However, only a few works on the variable selection in GPRM were done. In this paper, we propose a new algorithm to build a good prediction model among some GPRMs. It is a post-work of the algorithm that includes the Welch method suggested by previous researchers. The proposed algorithms select some non-zero regression coefficients (${\beta}^{\prime}s$) using forward and backward methods along with the Lasso guided approach. During this process, the fixed were covariance parameters (${\theta}^{\prime}s$) that were pre-selected by the Welch algorithm. We illustrated the superiority of our proposed models over the Welch method and non-selection models using four test functions and one real data example. Future extensions are also discussed.

• Fashion & Textile Research Journal
• /
• v.17 no.2
• /
• pp.278-286
• /
• 2015

Fashion industry has tried to adopt the virtual garment technology to reduce the time and effort spent on sample creation. For garment manufacturers to adopt the virtual garment technology as an alternative to sample creation, 3D avatars that meet the needs of each brand should be developed. Virtual garment softwares that are available in the market provide avatars with standardized body models and allow to modify the size by manually entering size specifications. This study proposed a methodology to develop size specifications for 3D avatars as well as brand-customized production sizes. For this, a man's fashion brand which is using virtual garment technology is selected. And the Size Korea database is used to develop size specification based on the customers' body shape. This study developed regression equations on body size specifications, which in turn proposed a regression model to proportionately change size specifications of 3D fitting-models. Based on the each body size calculated by the regression model, a standard model is created, and the skeleton-skin algorithm is applied to the regression model to obtain the results of size changes. Then, the 3D model sizes are tested for size changes as well as measured, which verifies that the regression model reflects body size changes.

• Journal of the Korean Solar Energy Society
• /
• v.40 no.2
• /
• pp.37-48
• /
• 2020

This study compares cloud radiation model (CRM) and sunshine fraction radiation model (SFRM) according to the solar altitude using hourly sunshine duration (SD) and cloud cover (CC) data. Solar irradiance measurements are not easy for the expensive measuring equipment and precise measuring technology. The two models with the site fitting and South Korea coefficients have been analyzed for fourteen cities of South Korea during the period (1986-2015) and evaluated using the root mean square error (RMSE) and the mean bias error (MBE). From the comparison of the results, it is found that the SFRM with the site fitting coefficients could be the best method for fourteen locations. It may be concluded that the SFRM models of South Korea coefficients generated in this study may be used reasonably well for calculating the hourly horizontal global irradiance (HGI) at any other location of South Korea.

• Journal of the Korean GEO-environmental Society
• /
• v.16 no.12
• /
• pp.23-35
• /
• 2015

Nonlinear soil behavior before failure under dynamic loading is often implemented in a numerical analysis code by a mathematical fitting function model with Masing's rule. However, the model may show different behavior with an experimental results obtained from laboratory test in damping ratio corresponding secant shear modulus for a certain shear strain rage. The difference may come from an unique soil characteristics which is unable to implement by using the existing mathematical fitting model. As of now, several fitting models have been suggested to overcome the difference between model and real soil behavior but consequence of the difference in dynamic analysis is not reviewed yet. In this paper, the effect of the difference on site response was examined through nonlinear response history analysis. The analysis was verified and calibrated with well defined dynamic geotechnical centrifuge test. Site response analyses were performed with three mathematical fitting function models and compared with the centrifuge test results in prototype scale. The errors on peak ground acceleration between analysis and experiment getting increased as increasing the intensity of the input motion. In practical point of view, the analysis results of accuracy with the fitting model is not significant in low to mid input motion intensity.

• The Korean Journal of Nuclear Medicine
• /
• v.39 no.4
• /
• pp.257-262
• /
• 2005

Purpose: Gastric emptying scan (GES) is usually acquired up to 2 hours. Our study investigated whether a fraction of meal-retention in the stomach at 120 minutes (FR120) was predicted from the data measured for 90 minutes by using non-linear curve fitting. We aimed at saving the delayed imaging by utilizing mathematical models. Materials and Methods: Ninety-six patients underwent GES immediately after taking a boiled egg with 74 MBq (2 mCi) Tc-99m DTPA. The patients were divided into Group I ($T_{1/2}\;{\leq}90\;min$) and Group II ($90\;min). Group I (n=51) had 21 men and 30 women, and Group II (n=45) 15 men and 30 women. There was no significant difference in age and sex between the two groups. Simple exponential, power exponential, and modified power exponential curves were acquired from the measured fraction of meal-retention at each time (0, 15, 30, 45, 60, 75, and 90 min) by non-linear curve fitting ($MATLAB^{\circledR}$ 5.3) and another simple exponential fitting was performed on the fractions at late times (60, 75, and 90 min). A predicted FR120 was calculated from the acquired functional formulas. A correlation coefficient between the measured FR120 and the predicted FR120 was computed ($MedCalc^{\circledR}$ 6.0). Results: Correlation coefficients(r) between the measured FR120 and the predicted FR120 of each mathematical functions were as follows: simple exponential function (Group I: 0.8558, Group II: 0.5982, p<0.0001), power exponential function (Group I: 0.8755, Group II: 0.6008, p<0.0001), modified power exponential function (Group I: 0.8892, Group II: 0.5882, p<0.0001), and simple exponential function at the late times(Group I: 0.9085, Group II: 0.6832, p<0.0001). In all the fitting models, the predicted FR120 were significantly correlated with the measured FR120 in Group I but not in Group II. There was no statistically significant difference in correlation among the 4 mathematical models. Conclusion: In the cases with $T_{1/2}\;{\leq}90\;min$, the predicted FR120 is significantly correlated with the measured FR120. Therefore, FR120 can be predicted from the data measured for 90 minutes by using non-linear curve fitting, saving the delayed imaging after 90 minutes when $T_{1/2}\;{\leq}90\;min$ is ascertained.

• Korea-Australia Rheology Journal
• /
• v.20 no.4
• /
• pp.197-211
• /
• 2008

The purpose of this study is mainly directed towards present of viewpoints on critical and commentary analysis on blood rheology, blood viscosity models, and physiological flow conditions. Understanding these basics is fundamental to meet the need for a sufficient and reliable CFD model of blood. Most of the used viscosity models on this manner have determined from parameter fitting on experimental viscosity data. Availability of experimental data from literature to define viscosity models of CFD analysis should be accurately chosen and treated in order to avoid any errors. Several basic gaps that limit the CFD model results are identified and given opportunities for future research.

• Communications for Statistical Applications and Methods
• /
• v.19 no.6
• /
• pp.885-898
• /
• 2012

In this paper, we consider Bayesian approaches to partially linear models, in which a regression function is represented by a semiparametric additive form of a parametric linear regression function and a nonparametric regression function. We make a comparative study on the performance of widely used Bayesian partially linear models in terms of empirical analysis. Specifically, we deal with three Bayesian methods to estimate the nonparametric regression function, one method using Fourier series representation, the other method based on Gaussian process regression approach, and the third method based on the smoothness of the function and differencing. We compare the numerical performance of three methods by the root mean squared error(RMSE). For empirical analysis, we consider synthetic data with simulation studies and real data application by fitting each of them with three Bayesian methods and comparing the RMSEs.

• Bulletin of the Korean Chemical Society
• /
• v.35 no.10
• /
• pp.2906-2910
• /
• 2014

This report introduces Gaussian electrostatic models (GEMs) to account for charge penetration effects in water-water interactions, allowing electrostatic interactions to be accurately described. Three different Gaussian electrostatic models, GEM-3S, GEM-5S, and GEM-6S are designed with s-type Gaussian functions. The coefficients and exponents of the Gaussian functions are optimized using the electrostatic potential (ESP) fitting procedure based on that of the MP2/aug-cc-pVTZ method. The electrostatic energies of ten different water dimers that were calculated with GEM-6S agree well with the results of symmetry-adapted perturbation theory (SAPT), indicating that this designed model can be effectively applied to future water models.

• Korean Journal of Computational Design and Engineering
• /
• v.14 no.5
• /
• pp.346-354
• /
• 2009

BRDF (bidirectional reflectance distribution function) is critical in realistic simulation of material appearances since it models the directional characteristics of reflection of light. Although many BRDF models have been proposed so far, it is still not easy to find one specific model that could represent all the reflection properties of real materials such as generalized diffusion, off-specular reflection, Fresnel effect, and back scattering. In this paper, we compare three BRDF models including B-spline volume BRDF (BVB), Cook-Torrance, and Lafortune in their ability to represent the measured BRDF data for physically-based rendering. We show that B-spline volume BRDF surpass the others in quality of data fitting and rendering, especially for materials without specular reflections.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.27 no.3
• /
• pp.180-193
• /
• 2023

Approximating the implied volatilities and estimating the model parameters are important topics in quantitative finance. This study proposes an approximation formula for short-maturity near-the-money implied volatilities in stochastic volatility models. A general second-order nonlinear PDE for implied volatility is derived in terms of time-to-maturity and log-moneyness from the Feyman-Kac formula. Using regularity conditions and the Taylor expansion, an approximation formula for implied volatility is obtained for short-maturity nearthe-money call options in two stochastic volatility models: Heston model and SABR model. In addition, we proposed a novel numerical method to estimate model parameters. This method reduces the number of model parameters that should be estimated. Generating sample data on log-moneyness, time-to-maturity, and implied volatility, we estimate the model parameters fitting the sample data in the above two models. Our method provides parameter estimates that are close to true values.