• Communications for Statistical Applications and Methods
• /
• v.22 no.6
• /
• pp.665-674
• /
• 2015

Lack of a general matrix formula hampers implementation of the semi-partial correlation, also known as part correlation, to the higher-order coefficient. This is because the higher-order semi-partial correlation calculation using a recursive formula requires an enormous number of recursive calculations to obtain the correlation coefficients. To resolve this difficulty, we derive a general matrix formula of the semi-partial correlation for fast computation. The semi-partial correlations are then implemented on an R package ppcor along with the partial correlation. Owing to the general matrix formulas, users can readily calculate the coefficients of both partial and semi-partial correlations without computational burden. The package ppcor further provides users with the level of the statistical significance with its test statistic.

• Communications for Statistical Applications and Methods
• /
• v.22 no.6
• /
• pp.675-683
• /
• 2015

In the 21st century, we now face the serious problems of the enormous consumption of the energy resources. Depending on the power consumption increases, both China and South Korea face a reduction in available resources. This paper considers the regression models and time-series models to compare the performance of the forecasting accuracy based on Mean Absolute Percentage Error (MAPE) in order to forecast the electricity demand accurately on the short-term period (68 months) data in Northeast China and find the relationship with Korea. Among the models the support vector regression (SVR) model shows superior performance than time-series models for the short-term period data and the time-series models show similar results with the SVR model when we use long-term period data.

• Communications for Statistical Applications and Methods
• /
• v.22 no.6
• /
• pp.531-542
• /
• 2015

In this survey, we use the normal linear model to demonstrate the use of the linear Bayes method. The superiorities of linear Bayes estimator (LBE) over the classical UMVUE and MLE are established in terms of the mean squared error matrix (MSEM) criterion. Compared with the usual Bayes estimator (obtained by the MCMC method) the proposed LBE is simple and easy to use with numerical results presented to illustrate its performance. We also examine the applications of linear Bayes method to some other distributions including two-parameter exponential family, uniform distribution and inverse Gaussian distribution, and finally make some remarks.

• Communications for Statistical Applications and Methods
• /
• v.22 no.6
• /
• pp.543-556
• /
• 2015

We present a survey of contributions that defined the nature and extent of robust statistics for the last 50 years. From the pioneering work of Tukey, Huber, and Hampel that focused on robust location parameter estimation, we presented various generalizations of these estimation procedures that cover a wide variety of models and data analysis methods. Among these extensions, we present linear models, clustered and dependent observations, times series data, binary and discrete data, models for spatial data, nonparametric methods, and forward search methods for outliers. We also present the current interest in robust statistics and conclude with suggestions on the possible future direction of this area for statistical science.

• Communications for Statistical Applications and Methods
• /
• v.21 no.3
• /
• pp.261-274
• /
• 2014

In this paper we develop Bayes and empirical Bayes estimators of the finite population mean with the assumption of posterior linearity rather than normality of the superpopulation under the balanced loss function. We compare the performance of the optimal Bayes estimator with ones of the classical sample mean and the usual Bayes estimator under the squared error loss with respect to the posterior expected losses, risks and Bayes risks when the underlying distribution is normal as well as when they are binomial and Poisson.

• Communications for Statistical Applications and Methods
• /
• v.21 no.3
• /
• pp.235-243
• /
• 2014

Constrained Bayesian estimates overcome the over shrinkness toward the mean which usual Bayes and empirical Bayes estimates produce by matching first and second empirical moments; subsequently, a constrained Bayes estimate is recommended to use in case the research objective is to produce a histogram of the estimates considering the location and dispersion. The well-known squared error loss function exclusively emphasizes the precision of estimation and may lead to biased estimators. Thus, the balanced loss function is suggested to reflect both goodness of fit and precision of estimation. In insurance pricing, the accurate location estimates of risk and also dispersion estimates of each risk group should be considered under proper loss function. In this paper, by applying these two ideas, the benefit of the constrained Bayes estimates and balanced loss function will be discussed; in addition, application effectiveness will be proved through an analysis of real insurance accident data.

• Communications for Statistical Applications and Methods
• /
• v.21 no.3
• /
• pp.245-251
• /
• 2014

Two exponential inequalities are established for a wide class of general weakly dependent sequences of random variables, called ${\psi}$-weakly dependent process which unify weak dependence conditions such as mixing, association, Gaussian sequences and Bernoulli shifts. The ${\psi}$-weakly dependent process includes, for examples, stationary ARMA processes, bilinear processes, and threshold autoregressive processes, and includes essentially all classes of weakly dependent stationary processes of interest in statistics under natural conditions on the process parameters. The two exponential inequalities are established on more general conditions than some existing ones, and are proven in simpler ways.

• Communications for Statistical Applications and Methods
• /
• v.21 no.3
• /
• pp.225-234
• /
• 2014

We propose a novel hierarchical clustering for distribution valued dissimilarities. Analysis of large and complex data has attracted significant interest. Symbolic Data Analysis (SDA) was proposed by Diday in 1980's, which provides a new framework for statistical analysis. In SDA, we analyze an object with internal variation, including an interval, a histogram and a distribution, called a symbolic object. In the study, we focus on a cluster analysis for distribution valued dissimilarities, one of the symbolic objects. A hierarchical clustering has two steps in general: find out step and update step. In the find out step, we find the nearest pair of clusters. We extend it for distribution valued dissimilarities, introducing a measure on their order relations. In the update step, dissimilarities between clusters are redefined by mixture of distributions with a mixing ratio. We show an actual example of the proposed method and a simulation study.

• Communications for Statistical Applications and Methods
• /
• v.21 no.3
• /
• pp.213-224
• /
• 2014

In this paper, we study the problem of transforming to normality. We propose to estimate the transformation parameter by minimizing a weighted squared distance between the empirical characteristic function of transformed data and the characteristic function of the normal distribution. Our approach also allows for other symmetric target characteristic functions. Asymptotics are established for a random sample selected from an unknown distribution. The proofs show that the weight function $t^{-2}$ needs to be modified to have thinner tails. We also propose the method to compute the influence function for M-equation taking the form of U-statistics. The influence function calculations and a small Monte Carlo simulation show that our estimates are less sensitive to a few outliers than the maximum likelihood estimates.

• Communications for Statistical Applications and Methods
• /
• v.20 no.2
• /
• pp.115-127
• /
• 2013

This paper provides an effective stock portfolio screening tool for socially responsible investment (SRI) based upon corporate social responsibility (CSR) and financial performance. The proposed approach utilizes nonparametric frontier models. Data envelopment analysis (DEA) has been used to build SRI portfolios in a few previous works; however, we show that free disposal hull (FDH), a similar model that does not assume the convexity of the technology, yields superior results when applied to a stock universe of 253 Korean companies. Over a four-year time span (from 2006 to 2009) the portfolios selected by the proposed method consistently outperform those selected by DEA as well as the benchmark.