• Communications for Statistical Applications and Methods
• /
• v.29 no.4
• /
• pp.471-486
• /
• 2022

Autoregressive moving average (ARMA) models involve nonlinearity in the model coefficients because of unobserved lagged errors, which complicates the likelihood function and makes the posterior density analytically intractable. In order to overcome this problem of posterior analysis, some approximation methods have been proposed in literature. In this paper we first review the main analytic approximations proposed to approximate the posterior density of ARMA models to be analytically tractable, which include Newbold, Zellner-Reynolds, and Broemeling-Shaarawy approximations. We then use the Kullback-Leibler divergence to study the relation between these three analytic approximations and to measure the distance between their derived approximate posteriors for ARMA models. In addition, we evaluate the impact of the approximate posteriors distance in Bayesian estimates of mean and precision of the model coefficients by generating a large number of Monte Carlo simulations from the approximate posteriors. Simulation study results show that the approximate posteriors of Newbold and Zellner-Reynolds are very close to each other, and their estimates have higher precision compared to those of Broemeling-Shaarawy approximation. Same results are obtained from the application to real-world time series datasets.

• Journal of Animal Science and Technology
• /
• v.45 no.3
• /
• pp.343-354
• /
• 2003

Simulation studies were carried out to investigate the responses of selection for three carcass traits (longissimus muscle area: EMA, fat thickness: BF, and marbling score: MS) based on either adjusted phenotypes (APH) or estimated breeding values (EBV) in multivariate animal model with different breeding schemes. Selection responses were estimated and compared on six different models with respect to breeding schemes using either carcass measurements or real time ultrasonic (RTU) scans generated by Monte Carlo computer simulation supporting closed breeding population. From the base population with 100 sires and 2000 dams, 20 sires and 1000 dams by each generation were selected by either APH or EBV for 10 generations. Relative economic weights were equal of three traits as EMA(1): BF(－1) : MS(1) for standardized either APH or EBV. For first two models which were similarly designed with current progeny-test program in Korean cattle, three carcass traits with records either only on male progenies (Model 1) or on male and female progenies (Model 2) were used for selecting breeding stocks. Subsequently, generation intervals on males were assumed as 6${\sim}$10 years in these two models. The other two models were designed with tools of selection by RTU rather than carcass measurements with genetic correlations of 0.81${\sim}$0.97 between RTU and corresponding carcass traits in addition to whether with records (Model 4) or without records (Model 3) on female. In these cases, generation intervals on males were assumed as 2${\sim}$4 years. The remaining last two models were designed as similar with Models 3 and 4 except genetic correlations of 0.63${\sim}$0.68 between RTU and corresponding carcass traits with records (Model 6) and without records (Model 5) on females. The results from 10 replicates on each model and selecting methods suggested that responses indirect selection for carcass traits in Model 4 were 1.66${\sim}$2.44 times efficient rather than those in Model 1. Otherwise, in Model 6 with assuming moderate genetic correlations, those efficiencies were 1.18${\sim}$2.08 times with comparing to responses in Model 1. However, selection response for marbling score was the smallest among three carcass traits because of small variation of measurements. From these results, this study suggested that indirect selection using RTU technology for improving high meat qualities in Korean cattle would be valuable with modifying measuring rules of marbling score forward to large variation or modifying relative economic weight for selection.

• KDI Journal of Economic Policy
• /
• v.26 no.2
• /
• pp.3-62
• /
• 2004

This paper estimates Korea's natural rate of unemployment using various estimation methods such as pure time-series methods, reduced-form methods, and structural form methods, with discussion about relative advantages and disadvantages of each estimation method. This paper also provides the confidence interval of the estimated natural rate of unemployment by the Monte Carlo integration method. Though multivariate unobserved component model exhibits better performance in many aspects than other estimation methods, awareness should be raised for a potential misspecification problem of a multivariate unobserved component model. Considering that each method has its own advantages and disadvantages, it is recommended to make an inference on the natural rate of unemployment based on common results among various methods. Korea's natural rate of unemployment was estimated to be around 3.8~4.0% on average in the period of 1979:I~1987:IV, and to decline to 2.5~2.9% in the period of 1988:I~1997:IV. During the Asian crisis, it is estimated to peak at near 4.8% and to have been on a downward trend since then.

• Communications for Statistical Applications and Methods
• /
• v.18 no.1
• /
• pp.57-69
• /
• 2011

Researchers are continuously trying to find innovative diagnostic tests and published articles are accumulating at an enormous rate in many medical fields. Meta-analysis enables previously published study results to be reviewed and summarized; therefore, an objective assessment of diagnostic tests can be done with a meta-analysis of sensitivities and specificities. Data obtained by applying two diagnostic tests to a well-defined group of diseased patients produce a pair of sensitivity and by applying the same medical tests to a group of non-diseased subjects produce a pair of specificity. The statistical tests in the meta-analysis need to consider the correlatedness of the results from two diagnostic tests applied to the same diseased and non-diseased subjects. The associations between two diagnostic test results are often found to be unequal for the diseased and non-diseased subjects. In this paper, multivariate meta-analytic methods are studied by taking into account the different associations between correlated variables. On the basis of Monte Carlo simulations, we evaluate the performance of the multivariate meta-analysis methods proposed in this paper.

• The Korean Journal of Applied Statistics
• /
• v.31 no.4
• /
• pp.507-516
• /
• 2018

Widely used among the various single imputation methods is k-nearest neighbors (KNN) imputation due to its robustness even when a parametric model such as multivariate normality is not satisfied. We propose a weighted adaptive nearest neighbors imputation method that combines the adaptive nearest neighbors imputation method that accounts for the local features of the data in the KNN imputation method and weighted k-nearest neighbors method that are less sensitive to extreme value or outlier among k-nearest neighbors. We conducted a Monte Carlo simulation study to compare the performance of the proposed imputation method with previous imputation methods.

• The Korean Journal of Applied Statistics
• /
• v.15 no.2
• /
• pp.297-310
• /
• 2002

Liang and Zeger proposed generalized estimating equations(GEE) for analyzing repeated data which is discrete or continuous. GEE model can be extended to model for repeated categorical data and its estimator has asymptotic multivariate normal distribution in large sample sizes. But GEE is based on large sample asymptotic theory. In this paper, we study the properties of GEE estimators for repeated ordinal data in small sample sizes. We generate ordinal repeated measurements for two groups using two methods. Through Monte Carlo simulation studies we investigate the empirical type 1 error rates, powers, relative efficiencies of the GEE estimators, the effect of unequal sample size of two groups, and the performance of variance estimators for polytomous ordinal response variables, especially in small sample sizes.

• Journal of Electrical Engineering and Technology
• /
• v.9 no.2
• /
• pp.461-470
• /
• 2014

This paper presents a probabilistic power flow (PPF) analysis method for distribution network incorporating the randomness and correlation of photovoltaic (PV) generation. Based on the multivariate kernel density estimation theory, the probabilistic model of PV generation is proposed without any assumption of theoretical parametric distribution, which can accurately capture not only the randomness but also the correlation of PV resources at adjacent locations. The PPF method is developed by combining the proposed PV model and Monte Carlo technique to evaluate the influence of the randomness and correlation of PV generation on the performance of distribution networks. The historical power output data of three neighboring PV generators in Oregon, USA, and 34-bus/69-bus radial distribution networks are used to demonstrate the correctness, effectiveness, and application of the proposed PV model and PPF method.

• The Korean Journal of Applied Statistics
• /
• v.22 no.6
• /
• pp.1355-1364
• /
• 2009

This work propose a robust test for location parameters in multivariate data based on MVE and MCD with the affine equivariance and the high-breakdown properties. We consider the hypothesis testing satisfying high efficiency and high test power simultaneously to bring in the one-step reweighting procedure upon high-breakdown estimators, which generally suffer from the low efficiency and, as a result, usually used only in the exploratory analysis. Monte Carlo study shows that the suggested method retains nominal significance levels and higher testing power without regard to various population distributions than a Hotelling's $T^2$ test. In an example, a data set containing known outliers does not make an influence toward our proposal, while it renders a Hotelling's $T^2$ useless.

• Journal of the Society of Naval Architects of Korea
• /
• v.57 no.2
• /
• pp.80-87
• /
• 2020

As the Arctic sea ice decreases due to various reasons such as global warming, the demand for ships and offshore structures operating in the Arctic region is steadily increasing. In the case of sea ice, the anisotropy is caused by the uncertainty inside the material. For most of the research, nevertheless, estimating the ice load has been treated deterministically. With regard to this, in this paper, a four-point bending strength analysis of an ice specimen was attempted using a stochastic finite element method. First, spatial distribution of the material properties used in the yield criterion was assumed to be a multivariate Gaussian random field. After that, a direct method, which is a sort of stochastic finite element method, and a sensitivity method using the sensitivity of response for random variables were proposed for calculating the probabilistic distribution of ice specimen strength. A parametric study was conducted with different mean vectors and correlation lengths for each material property used in the above procedure. The calculation time was about ten seconds for the direct method and about three minutes for the sensitivity methods. As the cohesion and correlation length increased, the mean value of the critical load and the standard deviation increased. On the contrary, they decreased as the friction angle increased. Also, in all cases, the direct and sensitivity methods yielded very similar results.

• The Korean Journal of Applied Statistics
• /
• v.11 no.2
• /
• pp.435-449
• /
• 1998

In this article we have introduced a Bayes criterion for the variable selection in multiple discriminant analysis (MDA). The criterion is a default Bayes factor for the comparision of homo/heteroscadasticity of the multivariate normal means. The default Bayes factor is obtained from a development of the imaginary training sample method introduced by Spiegelhalter and Smith (1982). Based an the criterion, we also provided a test for additional discrimination in MDA. The advantage of the criterion is that it is not only applicable for the optimal subset selection method but for the stepwise method. More over, the criterion can be reduced to that for two-group discriminant analysis. Thus the criterion can be regarded as an unified alternative to variable selection criteria suggested by various sampling theory approaches. To illustrate the performance of the criterion, a numerical study has bean done via Monte Carlo experiment.