• Journal of the Korean Data and Information Science Society
• /
• v.27 no.6
• /
• pp.1661-1671
• /
• 2016

Risk analysis is a systematic study of uncertainties and risks we encounter in business, engineering, public policy, and many other areas. Value at Risk (VaR) is one of the most widely used risk measurements in risk management. In this paper, the Korean Composite Stock Price Index data has been utilized to model the VaR employing the classical ARMA (1,1)-GARCH (1,1) models with normal, t, generalized hyperbolic, and generalized pareto distributed errors. The aim of this paper is to compare the performance of each model in estimating the VaR. The performance of models were compared in terms of the number of VaR violations and Kupiec exceedance test. The GARCH-GPD likelihood ratio unconditional test statistic has been found to have the smallest value among the models.

• Journal of the Korean Statistical Society
• /
• v.28 no.3
• /
• pp.371-387
• /
• 1999

For possibly nonstationary autoregressive moving average, modeling based on the original observations rather than the differenced observations is considered. Under this scheme, sample autocorrelation functions, parameter estimates, model diagnostic statistics, and prediction are all computed from the original data instead of the differenced data. The methods and results established under stationarity of data are shown to naturally extend to the nonstationarity of one autoregressive unit root. The sample ACF and PACF can be used for ARMA order determination. The BIC order is strongly consistent. The parameter estimates are asymptotically normal. The portmanteau statistic has chi-square distribution. The predictor is asymptotically equivalent to that based on the differenced data.

• Communications for Statistical Applications and Methods
• /
• v.23 no.5
• /
• pp.399-409
• /
• 2016

Many of studies have suggested the modifications on Weibull distribution to model the non-monotone hazards. In this paper, we combine two cumulative hazard functions and propose a new modified Weibull distribution function. The newly suggested distribution will be named as a new flexible Weibull distribution. Corresponding hazard function of the proposed distribution shows flexible (monotone or non-monotone) shapes. We study the characteristics of the proposed distribution that includes ageing behavior, moment, and order statistic. We also discuss an estimation method for its parameters. The performance of the proposed distribution is compared with existing modified Weibull distributions using various types of hazard functions. We also use real data example to illustrate the efficiency of the proposed distribution.

• Journal of the Korean Statistical Society
• /
• v.13 no.1
• /
• pp.48-56
• /
• 1984

This paper considers one-sample and two-sample test for the logistic function by means of Kolmororov-Smirnov type statistics. The standard tables used for the Kolmogorov-Smirnov test are valid only when the function is completely specified; but they are not valid if the parameters of function are estimated from the sample. This note presents modified tables for the Kolmogorov-Sminov type staistic. These tables can be used to test the hypothesis that a sample comes from a logistic function when shape parameter $(\alpha)$ and location parameter $(\beta)$ must be estimated from the sample by the method of maximum likelihood. Monte Carlo method is employed to calculate the criticla values of the test. The tables of the critical values are provided.

• Communications for Statistical Applications and Methods
• /
• v.5 no.3
• /
• pp.847-853
• /
• 1998

Suppose we want to compare following non-hierarchical log-linear models, $H_0:f(x, heta inTheta_a)$ vs H_1:g(x, heta inTheta_eta); for; Theta_a,;Theta_etasubsetTheta;such;that;Theta_$\alpha$/ Theta_eta$. The goodness of fit test using the likelihood ratio test statistic for comparing these models could not be acceptable. By using the polyhedrons plots of Choi and Hong (1995), we propose a method to decide a better model between two non-hierarchical log-linear models $f(x: heta inTheta_a) and g(x: heta inTheta_eta)$.

• Animal Systematics, Evolution and Diversity
• /
• v.30 no.1
• /
• pp.39-43
• /
• 2014

In this brief report, alternative time-varying diversification rate models were fitted onto the phylogeny of global amphibians by considering one-constant-rate (OCR), one-continuous-shift (OCS) and multiple-discrete- shifts (MDS) situations. The OCS diversification model was rejected by ${\gamma}$ statistic (${\gamma}=-5.556$, p<0.001), implying the existence of shifting diversification rates for global amphibian phylogeny. Through model selection, MDS diversification model outperformed OCS and OCR models using "laser" package under R environment. Moreover, MDS models, implemented using another R package "MEDUSA", indicated that there were sixteen shifts over the internal nodes for amphibian phylogeny. Conclusively, both OCS and MDS models are recommended to compare so as to better quantify rate-shifting trends of species diversification. MDS diversification models should be preferential for large phylogenies using "MEDUSA" package in which any arbitrary numbers of shifts are allowed to model.

• Communications for Statistical Applications and Methods
• /
• v.15 no.4
• /
• pp.605-614
• /
• 2008

Oh (2008) has proposed the one-sided likelihood ratio test of the equality of two agreement measures. However the use of this test may be limited since the computations of test statistic and critical value are not easy. We propose a test for comparing two dependent agreements using some well known tests for marginal homogeneity, for instance, Bhapkar test, Stuart-Maxwell test. Data obtained from 2008 world figure skating championship ladies single is analyzed for illustration purposes.

• The Korean Journal of Applied Statistics
• /
• v.15 no.1
• /
• pp.187-199
• /
• 2002

When the sample size is small, exact tests are often employed because the asymptotic distribution of the test statistic is in doubt. The advantage of exact tests is that it is guaranteed to bound the type I error probability to the nominal level. In this paper we review the methods of constructing exact tests, the algorithm and commercial software. We also examine the difference between exact p-values obtained from exact tests and true p-values obtained from the true underlying distribution.

• Journal of Integrative Natural Science
• /
• v.13 no.2
• /
• pp.76-82
• /
• 2020

If normality of an observed data is not a viable assumption, we can carry out normal-theory analyses by suitable transforming data. Power transformation by Box and Cox, one of the transformation methods, is derived the power which maximized the likelihood function. But it doesn't induces the closed form in mathematical analysis. In this paper, we compose some R the syntax of which is easier than other statistical packages for deriving the power with using numerical methods. Also, by using R, we show the transformed data approximately distributed the normal through Q-Q plot in univariate and bivariate cases with some examples. Finally, we present the value of a goodness-of-fit statistic(AD) and its p-value for normal distribution. In the similar procedure, this method can be extended to more than bivariate case.

• ETRI Journal
• /
• v.37 no.4
• /
• pp.736-742
• /
• 2015

A serial concatenated decoding algorithm with dynamic threshold is proposed for low-density parity-check codes with short and medium code lengths. The proposed approach uses a dynamic threshold to select a decoding result from belief propagation decoding and order statistic decoding, which improves the performance of the decoder at a negligible cost. Simulation results show that, under a high SNR region, the proposed concatenated decoder performs better than a serial concatenated decoder without threshold with an Eb/N0 gain of above 0.1 dB.