• The Korean Journal of Applied Statistics
• /
• v.35 no.2
• /
• pp.217-227
• /
• 2022

Quantile regression is widely used in many fields based on the advantage of providing an efficient tool for examining complex information latent in variables. However, modern large-scale and high-dimensional data makes it very difficult to estimate the quantile regression model due to limitations in terms of computation time and storage space. Divide-and-conquer is a technique that divide the entire data into several sub-datasets that are easy to calculate and then reconstruct the estimates of the entire data using only the summary statistics in each sub-datasets. In this paper, we studied on a variable selection method using Bayes information criteria by applying the divide-and-conquer technique to the penalized quantile regression. When the number of sub-datasets is properly selected, the proposed method is efficient in terms of computational speed, providing consistent results in terms of variable selection as long as classical quantile regression estimates calculated with the entire data. The advantages of the proposed method were confirmed through simulation data and real data analysis.

• Journal of the Korean Statistical Society
• /
• v.34 no.4
• /
• pp.331-344
• /
• 2005

Inference on current data could be more reliable if there exist similar data based on previous studies. Ibrahim and Chen (2000) utilize these data to characterize the power prior. The power prior is constructed by raising the likelihood function of the historical data to the power $a_o$, where $0\;{\le}\;a_o\;{\le}\;1$. The power prior is a useful informative prior in Bayesian inference. However, for model selection or model comparison problems, the propriety of the power prior is one of the critical issues. In this paper, we suggest two joint power priors for the power law process and show that they are proper under some conditions. We demonstrate our results with a real dataset and some simulated datasets.

• The Korean Journal of Applied Statistics
• /
• v.27 no.6
• /
• pp.1003-1016
• /
• 2014

Proper missing data imputation is an important procedure to obtain superior results for data analysis based on survey data. This paper deals with both a model based imputation method and model estimation method. We utilized a Bayesian method to solve a boundary solution problem in which we applied a maximum likelihood estimation method. We also deal with a missing mechanism model selection problem using forecasting results and a comparison between model accuracies. We utilized MWPE(modified within precinct error) (Bautista et al., 2007) to measure prediction correctness. We applied proposed ML and Bayesian methods to the Korean presidential election exit poll data of 2012. Based on the analysis, the results under the missing at random mechanism showed superior prediction results than under the missing not at random mechanism.

• The Korean Journal of Applied Statistics
• /
• v.16 no.2
• /
• pp.365-376
• /
• 2003

In this paper we reviewed a variety of non-Gaussian time series models, and studied the model selection criteria such as AIC and BIC to select proper models. We also considered the likelihood ratio test and applied it to analysis of Polio data set.

• Environmental and Resource Economics Review
• /
• v.31 no.3
• /
• pp.393-417
• /
• 2022

Using hourly SMP data from 2016 to 2020, this paper measures the weekly realized volatility and investigates the main force of its determinants. To this end, we extend the Bayesian variable selection by incorporating the regime-switching model which identifies important variables among a large number of predictors by regimes. We find that the increase in coal and nuclear generation, as well as solar power, reinforce the SMP volatility in both high volatility and low volatility regime. In contrast the increase in gas generation and gas price decrease SMP volatility when SMP volatility is high. These results suggest that the expansion of renewable energy according to 2050 Carbon Neutrality or energy transition policies increases SMP volatility but the increase in the gas generation or reduction of coal generation might offset its impact.

• The Korean Journal of Applied Statistics
• /
• v.24 no.5
• /
• pp.951-961
• /
• 2011

In this paper, we propose a Bayesian inference using the Markov Chain Monte Carlo(MCMC) method for the zero inflated negative binomial(ZINB) regression model. The proposed model allows the regression model for zero inflation probability as well as the regression model for the mean of the dependent variable. This extends the work of Jang et al. (2010) to the fully defiend ZINB regression model. In addition, we apply the proposed method to a real data example, and compare the efficiency with the zero inflated Poisson model using the DIC. Since the DIC of the ZINB is smaller than that of the ZIP, the ZINB model shows superior performance over the ZIP model in zero inflated count data with overdispersion.

• 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.

• The Korean Journal of Applied Statistics
• /
• v.32 no.6
• /
• pp.909-922
• /
• 2019

In this paper, we study the adaptive least absolute shrinkage and selection operator (LASSO) for the unstable autoregressive (AR) model. To identify the existence of the unit root, we apply the adaptive LASSO to the augmented Dickey-Fuller regression model, not the original AR model. We illustrate our method with simulations and a real data analysis. Simulation results show that the adaptive LASSO obtained by minimizing the Bayesian information criterion selects the order of the autoregressive model as well as the degree of differencing with high accuracy.

• Journal of Institute of Control, Robotics and Systems
• /
• v.22 no.2
• /
• pp.104-109
• /
• 2016

In this paper, a localization method for multiple robots based on Bayesian inference is proposed when multiple robots adopting multi-RAT (Radio Access Technology) communications exist in cognitive radio networks. Multiple robots are separately defined by primary and secondary users as in conventional mobile communications system. In addition, the heterogeneous spectrum environment is considered in this paper. To improve the performance of localization for multiple robots, a realistic multiple primary user distribution is explained by using the probabilistic graphical model, and then we introduce the Gibbs sampler strategy based on Bayesian inference. In addition, the secondary user selection minimizing the value of GDOP (Geometric Dilution of Precision) is also proposed in order to overcome the limitations of localization accuracy with Gibbs sampling. Via the simulation results, we can show that the proposed localization method based on GDOP enhances the accuracy of localization for multiple robots. Furthermore, it can also be verified from the simulation results that localization performance is significantly improved with increasing number of observation samples when the GDOP is considered.

• The KIPS Transactions:PartD
• /
• v.10D no.5
• /
• pp.805-812
• /
• 2003

This paper presents a stochastic model for the software failure phenomenon based on a nonhomogeneous Poisson process (NHPP) and performs Bayesian inference using prior information. The failure process is analyzed to develop a suitable mean value function for the NHPP; expressions are given for several performance measure. The parametric inferences of the model using Logarithmic Poisson model, Crow model and Rayleigh model is discussed. Bayesian computation and model selection using the sum of squared errors. The numerical results of this models are applied to real software failure data. Tools of parameter inference was used method of Gibbs sampling and Metropolis algorithm. The numerical example by T1 data (Musa) was illustrated.