• Communications for Statistical Applications and Methods
• /
• v.30 no.2
• /
• pp.163-178
• /
• 2023

GARCH-X(1, 1) model specifies that conditional variance follows an AR(1) process and includes a past exogenous variable. This study proposes a new class from that model by allowing a more general (non-linear) variance function to follow an AR(1) process. The functions applied to the variance equation include exponential, Tukey's ladder, and Yeo-Johnson transformations. In the framework of normal and student-t distributions for return errors, the empirical analysis focuses on two stock indices data in developed countries (FTSE100 and SP500) over the daily period from January 2000 to December 2020. This study uses 10-minute realized volatility as the exogenous component. The parameters of considered models are estimated using the adaptive random walk metropolis method in the Monte Carlo Markov chain algorithm and implemented in the Matlab program. The 95% highest posterior density intervals show that the three transformations are significant for the GARCHX(1, 1) model. In general, based on the Akaike information criterion, the GARCH-X(1, 1) model that has return errors with student-t distribution and variance transformed by Tukey's ladder function provides the best data fit. In forecasting value-at-risk with the 95% confidence level, the Christoffersen's independence test suggest that non-linear models is the most suitable for modeling return data, especially model with the Tukey's ladder transformation.

• Journal of information and communication convergence engineering
• /
• v.22 no.2
• /
• pp.165-171
• /
• 2024

In this study, we present a novel approach for enhancing chest X-ray image classification (normal, Covid-19, edema, mass nodules, and pneumothorax) by combining contrastive learning and machine learning algorithms. A vast amount of unlabeled data was leveraged to learn representations so that data efficiency is improved as a means of addressing the limited availability of labeled data in X-ray images. Our approach involves training classification algorithms using the extracted features from a linear fine-tuned Momentum Contrast (MoCo) model. The MoCo architecture with a Resnet34, Resnet50, or Resnet101 backbone is trained to learn features from unlabeled data. Instead of only fine-tuning the linear classifier layer on the MoCopretrained model, we propose training nonlinear classifiers as substitutes for softmax in deep networks. The empirical results show that while the linear fine-tuned ImageNet-pretrained models achieved the highest accuracy of only 82.9% and the linear fine-tuned MoCo-pretrained models an increased highest accuracy of 84.8%, our proposed method offered a significant improvement and achieved the highest accuracy of 87.9%.

• The Korean Journal of Applied Statistics
• /
• v.31 no.3
• /
• pp.367-382
• /
• 2018

Prediction models for a corporate bond rating in existing studies have been developed using various models such as linear regression, ordered logit, and random forest. Financial characteristics help build prediction models that are expected to be contained in the assigning model of the bond rating agencies. However, the ranges of bond ratings in existing studies vary from 5 to 20 and the prediction models were developed with samples in which the target companies and the observation periods are different. Thus, a simple comparison of the prediction accuracies in each study cannot determine the best prediction model. In order to conduct a fair comparison, this study has collected corporate bond ratings and financial characteristics from 2013 to 2017 and applied prediction models to them. In addition, we applied the elastic-net penalty for the linear regression, the ordered logit, and the ordered probit. Our comparison shows that data-driven variable selection using the elastic-net improves prediction accuracy in each corresponding model, and that the random forest is the most appropriate model in terms of prediction accuracy, which obtains 69.6% accuracy of the exact rating prediction on average from the 5-fold cross validation.

• Journal of the Korean Society for information Management
• /
• v.29 no.2
• /
• pp.193-204
• /
• 2012

The characteristics of citation and centrality measures in citation networks can be identified using multiple linear regression analyses. In this study, we examine the relationships between bibliometric indices and centrality measures in an article-level co-citation network to determine whether the linear model is the best fitting model and to suggest the necessity of data transformation in the analysis. 703 highly cited articles in Physics published in 2004 were sampled, and four indicators were developed as variables in this study: citation counts, degree centrality, closeness centrality, and betweenness centrality in the co-citation network. As a result, the relationship pattern between citation counts and degree centrality in a co-citation network fits a non-linear rather than linear model. Also, the relationship between degree and closeness centrality measures, or that between degree and betweenness centrality measures, can be better explained by non-linear models than by a linear model. It may be controversial, however, to choose non-linear models as the best-fitting for the relationship between closeness and betweenness centrality measures, as this result implies that data transformation may be a necessary step for inferential statistics.

• Atmosphere
• /
• v.25 no.4
• /
• pp.669-683
• /
• 2015

In this study, we investigated the prediction skills of four multiple linear regression methods for monthly air temperature over South Korea. We used simulation results from four regional climate models (RegCM4, SNURCM, WRF, and YSURSM) driven by two boundary conditions (NCEP/DOE Reanalysis 2 and ERA-Interim). We selected 15 years (1989~2003) as the training period and the last 5 years (2004~2008) as validation period. The four regression methods used in this study are as follows: 1) Homogeneous Multiple linear Regression (HMR), 2) Homogeneous Multiple linear Regression constraining the regression coefficients to be nonnegative (HMR+), 3) non-homogeneous multiple linear regression (EMOS; Ensemble Model Output Statistics), 4) EMOS with positive coefficients (EMOS+). It is same method as the third method except for constraining the coefficients to be nonnegative. The four regression methods showed similar prediction skills for the monthly air temperature over South Korea. However, the prediction skills of regression methods which don't constrain regression coefficients to be nonnegative are clearly impacted by the existence of outliers. Among the four multiple linear regression methods, HMR+ and EMOS+ methods showed the best skill during the validation period. HMR+ and EMOS+ methods showed a very similar performance in terms of the MAE and RMSE. Therefore, we recommend the HMR+ as the best method because of ease of development and applications.

• Journal of IKEEE
• /
• v.24 no.4
• /
• pp.1022-1027
• /
• 2020

In this paper, SVM-based speaker classification is experimented with GMM-supervector. To create a speaker cluster, conventional speaker change detection is performed with the KL distance using the SNR-based weighting function. SVM-based speaker classification consists of two steps. In the first step, SVM-based classification between UBM and speaker models is performed, speaker information is indexed in each cluster, and then grouped by speaker. In the second step, the SVM-based classification between UBM and speaker models is performed by inputting the speaker cluster group. Linear and RBF are applied as kernel functions for SVM-based classification. As a result, in the first step, the case of applying the linear kernel showed better performance than RBF with 148 speaker clusters, MDR 0, FAR 47.3, and ER 50.7. The second step experiment result also showed the best performance with 109 speaker clusters, MDR 1.3, FAR 28.4, and ER 32.1 when the linear kernel was applied.

• Wind and Structures
• /
• v.19 no.3
• /
• pp.233-247
• /
• 2014

Vortex-induced oscillation is a type of aeroelastic phenomenon, to which extended structures such as long-span bridges are most susceptible. The vortex-induced vibration (VIV) behaviors of a concerned bridge were investigated conventionally in virtue of wind tunnel tests on string-mounted sectional models. This necessitates the building of a linkage between the response of the sectional model and that of the prototype structure. Although many released literatures have related to this issue and provided suggestions, there is a lack of consistency among them. In this study, some theoretical models describing the vortex-induced structural motion, including the linear empirical model, the nonlinear empirical model and the modified (or generalized) nonlinear empirical model, are firstly reviewed. Then, the concept of equivalent mass density is introduced based on the principle that an equal input of energy should result in identical structural amplitudes. Based on these, the theoretical linkages between the amplitude of a section model and that corresponding to the prototype bridge are discussed with different analytical models. Theoretical derivation indicates that such connections are dependent mainly on two factors, one is the presupposed shape of deformation, and the other is the theoretical VIV model employed. The theoretical analysis in this study shows that, in comparison to the nonlinear empirical models, the linear one can result in obvious larger estimations of the full bridges' responses, especially in cases of cable-stayed bridges.

• Structural Engineering and Mechanics
• /
• v.46 no.2
• /
• pp.213-229
• /
• 2013

Although performance based assessment procedures are mainly developed for reinforced concrete and steel buildings, URM (Unreinforced Masonry) buildings occupy significant portion of buildings in earthquake prone areas of the world as well as in IRAN. Variability of material properties, non-engineered nature of the construction and difficulties in structural analysis of masonry walls make analysis of URM buildings challenging. Despite sophisticated finite element models satisfy the modeling requirements, extensive experimental data for definition of material behavior and high computational resources are needed. Recently, nonlinear equivalent frame models which are developed assigning lumped plastic hinges to isotropic and homogenous equivalent frame elements are used for nonlinear modeling of URM buildings. The equivalent frame models are not novel for the analysis of masonry structures, but the actual potentialities have not yet been completely studied, particularly for non-linear applications. In the present paper an effective tool for the non-linear static analysis of 2D masonry walls is presented. The work presented in this study is about performance assessment of unreinforced brick masonry buildings through nonlinear equivalent frame modeling technique. Reliability of the proposed models is tested with a reversed cyclic experiment conducted on a full scale, two-story URM building at the University of Pavia. The pushover curves were found to provide good agreement with the experimental backbone curves. Furthermore, the results of analysis show that EFM (Equivalent Frame Model) with Dolce RO (rigid offset zone) and shell element have good agreement with finite element software and experimental results.

• Earthquakes and Structures
• /
• v.12 no.5
• /
• pp.543-557
• /
• 2017

Results of an extensive study aiming to properly extend the well known pushover analysis into 3-D problems of asymmetric buildings are presented in this paper. The proposed procedure uses simple, 3 DOF, one-story models with shear-beam type elements in order to quantify the effects of inelastic torsional response of such buildings. Correction coefficients for the response quantities at the "stiff" and "flexible" sides are calculated using results from non-linear time history analyses of the simple models. Their values are then applied to the results of a simple, plane pushover analysis of the detailed building models. Results from the application of the new method for a set of three, conventionally designed, five-story buildings with high values of uniaxial eccentricities are compared with those obtained from multiple non-linear dynamic time history analyses, as well as from similar pushover methods addressing the same problem. This initial evaluation indicates that the proposed procedure is a clear improvement over the simple (conventional) pushover method and, in most cases, more accurate and reliable than the other methods considered. The accuracy, however, of all these methods is reduced substantially when they are applied to torsionally flexible buildings. Thus, for such challenging problems, use of inelastic dynamic analyses for a set of two component earthquake motions appears to be the preferable solution.

• Journal of Computing Science and Engineering
• /
• v.2 no.2
• /
• pp.200-219
• /
• 2008

The effective management of uncertainty is one of the most fundamental problems in medical decision making. According to the literatures review, most medical decision models rely on point estimates for input parameters. However, it is natural that they should be interested in the relationship between changes in those values and subsequent changes in model output. Therefore, the purpose of this study is to identify the ranges of numerical values for which each option will be most efficient with respect to the input parameters. The Nonhomogeneous Poisson Process(NHPP) was used for describing the behavior of aging chronic diseases. Three kind of failure models (linear, exponential, and power law) were considered, and each of these failure models was studied under the assumptions of unknown scale factor and known aging rate, known scale factor and unknown aging rate, and unknown scale factor and unknown aging rate, respectively. In addition, this study illustrated developed method with an analysis of data from a trial of immunotherapy in the treatment of chronic Granulomatous disease. Finally, the proposed design of Bayesian value of information analysis facilitates the effective use of the computing capability of computers and provides a systematic way to integrate the expert's opinions and the sampling information which will furnish decision makers with valuable support for quality medical decision making.