• Journal of the Korean Statistical Society
• /
• v.18 no.1
• /
• pp.46-61
• /
• 1989

This paper is concerned with the development of a nonparametric procedure for the statistical inference about the nonlinear regression parameters. A confidence region and a hypothesis testing procedure based on a class of signed linear rank statistics are proposed and the asymptotic distributions of the test statistic both under the null hypothesis and under a sequence of local alternatives are investigated. Some desirable asymptotic properties including the asymptotic relative efficiency are discussed for various score functions.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.17 no.7
• /
• pp.957-963
• /
• 2007

In this paper, Nonlinear Autoregressive (NAR) method based on Least Square-Support Vector Regression (LS-SVR) is introduced and tested for nonlinear sustained vowel modeling. In the database of total 43 sustained vowel of Benign Vocal Fold Lesions having aperiodic waveform, this nonlinear synthesizer near perfectly reproduced chaotic sustained vowels, and also conserved the naturalness of sound such as jitter, compared to Linear Predictive Coding does not keep these naturalness. However, the results of some phonation are quite different from the original sounds. These results are assumed that single-band model can not afford to control and decompose the high frequency components. Therefore multi-band model with wavelet filterbank is adopted for substituting single band model. As a results, multi-band model results in improved stability. Finally, nonlinear sustained vowel modeling using NAR based on LS-SVR can successfully reconstruct synthesized sounds nearly similar to original voiced sounds.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2002.05a
• /
• pp.541-546
• /
• 2002

• Steel and Composite Structures
• /
• v.37 no.2
• /
• pp.193-209
• /
• 2020

In this paper, the efficiency of five Machine Learning (ML) methods consisting of Deep Learning (DL), Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), and Gradient Tree Booting (GTB) for regression and classification of the Ultimate Load Factor (ULF) of nonlinear inelastic steel frames is compared. For this purpose, a two-story, a six-story, and a twenty-story space frame are considered. An advanced nonlinear inelastic analysis is carried out for the steel frames to generate datasets for the training of the considered ML methods. In each dataset, the input variables are the geometric features of W-sections and the output variable is the ULF of the frame. The comparison between the five ML methods is made in terms of the mean-squared-error (MSE) for the regression models and the accuracy for the classification models, respectively. Moreover, the ULF distribution curve is calculated for each frame and the strength failure probability is estimated. It is found that the GTB method has the best efficiency in both regression and classification of ULF regardless of the number of training samples and the space frames considered.

• Communications for Statistical Applications and Methods
• /
• v.27 no.6
• /
• pp.701-714
• /
• 2020

Quantitative high throughput screening (qHTS) assays are used to assess toxicity for many chemicals in a short period by collectively analyzing them at several concentrations. Data are routinely analyzed using nonlinear regression models; however, we propose a new method to analyze qHTS data using a nonlinear mixed effects model. qHTS data are generated by repeating the same experiment several times for each chemical; therefor, they can be viewed as if they are repeated measures data and hence analyzed using a nonlinear mixed effects model which accounts for both intra- and inter-individual variabilities. Furthermore, we apply a one-step approach incorporating robust estimation methods to estimate fixed effect parameters and the variance-covariance structure since outliers or influential observations are not uncommon in qHTS data. The toxicity of chemicals from a qHTS assay is classified based on the significance of a parameter related to the efficacy of the chemicals using the proposed method. We evaluate the performance of the proposed method in terms of power and false discovery rate using simulation studies comparing with one existing method. The proposed method is illustrated using a dataset obtained from the National Toxicology Program.

• Management Science and Financial Engineering
• /
• v.14 no.2
• /
• pp.89-104
• /
• 2008

Growth curves including Bass, Logistic and Gompertz functions are widely used in forecasting the market demand. Nonlinear least square method is often adopted for estimating the model parameters but it is difficult to set up the starting value for each parameter. If a wrong starting point is selected, the result may lead to erroneous forecasts. This paper proposes a method of selecting starting values for model parameters in estimating some growth curves by nonlinear least square method through grid search and transformation into linear regression model. Resealing the market data using the national economic index makes it possible to figure out the range of parameters and to utilize the grid search method. Application to some real data is also included, where the performance of our method is demonstrated.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.61 no.11
• /
• pp.1682-1688
• /
• 2012

This paper introduces a GP(Genetic Programming) based robust technique for temperature compensation in short-range prediction. Development of an efficient MOS(Model Output Statistics) is necessary to correct systematic errors of the model, because forecast models do not reliably determine weather conditions. Most of MOS use a linear regression to compensate a prediction model, therefore it is hard to manage an irregular nature of prediction. In order to solve the problem, a nonlinear and symbolic regression method using GP is suggested. The purpose of this study is to evaluate the accuracy of the estimation by a GP based nonlinear MOS for 3 days temperatures in Korean regions. This method is then compared to the UM model and has shown superior results. The training period of 2007-2009 summer is used, and the data of 2010 summer is adopted for verification.

• Communications for Statistical Applications and Methods
• /
• v.19 no.2
• /
• pp.293-301
• /
• 2012

Quantile regression proposed by Koenker and Bassett (1978) is a statistical technique that estimates conditional quantiles. The advantage of using quantile regression is the robustness in response to large outliers compared to ordinary least squares(OLS) regression. A regression tree approach has been applied to OLS problems to fit flexible models. Loh (2002) proposed the GUIDE algorithm that has a negligible selection bias and relatively low computational cost. Quantile regression can be regarded as an analogue of OLS, therefore it can also be applied to GUIDE regression tree method. Chaudhuri and Loh (2002) proposed a nonparametric quantile regression method that blends key features of piecewise polynomial quantile regression and tree-structured regression based on adaptive recursive partitioning. Lee and Lee (2006) investigated wage determinants in the Korean labor market using the Korean Labor and Income Panel Study(KLIPS). Following Lee and Lee, we fit three kinds of quantile regression tree models to KLIPS data with respect to the quantiles, 0.05, 0.2, 0.5, 0.8, and 0.95. Among the three models, multiple linear piecewise quantile regression model forms the shortest tree structure, while the piecewise constant quantile regression model has a deeper tree structure with more terminal nodes in general. Age, gender, marriage status, and education seem to be the determinants of the wage level throughout the quantiles; in addition, education experience appears as the important determinant of the wage level in the highly paid group.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.66 no.12
• /
• pp.1799-1803
• /
• 2017

This paper introduces the prediction methods of windstorm advisory using GP nonlinear compensation and SVM. The existing special report prediction is not specialized for strong wind, such as windstorm, because it is based on the wide range of predicted values for wind speed from low to high. In order to improve the performance of strong wind reporting prediction, a method that can efficiently classify boundaries of strong wind is necessary. First, evolutionary nonlinear regression based compensation technique is applied to obtain more accurate values of prediction for wind speed using UM data. Based on the prediction wind speed, the windstorm advisory is determined. Second, SVM method is applied to classify directly using the data of UM predictors and windstorm advisory. Above two methods are compared to evaluate of the performances for the windstorm data in Jeju Island in South Korea. The data of 2007-2009, 2011 year is used for training, and 2012 year is used for test.

• Journal of the Korean Data and Information Science Society
• /
• v.22 no.5
• /
• pp.999-1005
• /
• 2011

This paper proposes a weighted least squares support vector machine for asymmetric least squares regression. This method achieves nonlinear prediction power, while making no assumption on the underlying probability distributions. The cross validation function is introduced to choose optimal hyperparameters in the procedure. Experimental results are then presented which indicate the performance of the proposed model.