• Communications for Statistical Applications and Methods
• /
• v.10 no.1
• /
• pp.239-246
• /
• 2003

Statisticians, or data miners, are often requested to assess the importances of input variables in the given supervised learning model. For the purpose, one may rely on separate ad hoc measures depending on modeling types, such as linear regressions, the neural networks or trees. Consequently, the conceptual consistency in input variable importance measures is lacking, so that the measures cannot be directly used in comparing different types of models, which is often done in data mining processes, In this short communication, we propose a unified approach to the importance measurement of input variables. Our method uses sensitivity analysis which begins by perturbing the values of input variables and monitors the output change. Research scope is limited to the models for continuous output, although it is not difficult to extend the method to supervised learning models for categorical outcomes.

• Journal of the Korean Society of Safety
• /
• v.25 no.2
• /
• pp.89-94
• /
• 2010

This study deals with the traffic accident according to the improvement projects of frequent accident locations. The objective is to analyze the impact of improvements on the accident reduction. In pursuing the above, the study gives the particular attentions to developing the models based on the data of 70 intersections improved. The main results analyzed are as follows. First, 4 multiple linear regression accident models(total, side right-angle, rear end and side stripe accident) which were statistically significant were developed. Second, total accidents reduction by sight-distance and turning traffic flow improvements, side right-angle by sight-distance, over-speed and lane operation, rear end by turning traffic flow, signal and lane operation, and side stripe by traffic impedance improvements were analyzed. Finally, the above 4 models were evaluated to be statically significant through the correlation analysis and pair-sample t-test.

• Communications for Statistical Applications and Methods
• /
• v.29 no.3
• /
• pp.353-371
• /
• 2022

This paper analyzes death counts after World War II of several countries to identify and to compare their stochastic structures. The stochastic structures that this paper entertains are three structural time series models, a local level with a random walk model, a fixed local linear trend model and a local linear trend model. The structural time series models assume that a time series can be formulated directly with the unobserved components such as trend, slope, seasonal, cycle and daily effect. Random effect of each unobserved component is characterized by its own stochastic structure and a distribution of its irregular component. The structural time series models use the Kalman filter to estimate unknown parameters of a stochastic model, to predict future data, and to do filtering data. This paper identifies the best-fitted stochastic model for three types of death counts (Female, Male and Total) of each country. Two diagnostic procedures are used to check the validity of fitted models. Three criteria, AIC, BIC and SSPE are used to select the best-fitted valid stochastic model for each type of death counts of each country.

• Toxicological Research
• /
• v.18 no.4
• /
• pp.311-324
• /
• 2002

Toxicokinetic studies are intended to provide critical evaluation of drug disposition at toxico-logical doses and help understand the relationship between blood or tissue levels and the time course of toxic events. Relatively high dose levels wed in toxicokinetics, compared to pharmacokinetics, complicates absorption, protein binding, metabolism and elimination processes. In this mini review, frequently wed toxicokinetic models such as linear compartment models, physiological models, and nonlinear kinetic mod-ec are introduced. In addition, optimization of toxicokinetic studies, their role in the drug development process, and prediction oj human toxicokinetics based on animal data by interspecies scaling are briefly discussed.

• Communications for Statistical Applications and Methods
• /
• v.19 no.3
• /
• pp.423-432
• /
• 2012

This paper is concerned with the detection of multiple change-points in linear regression models. The proposed procedure relies on the local estimation for global change-point estimation. We propose a multiple change-point estimator based on the local least squares estimators for the regression coefficients and the split measure when the number of change-points is unknown. Its statistical properties are shown and its performance is assessed by simulations and real data applications.

• Communications for Statistical Applications and Methods
• /
• v.6 no.2
• /
• pp.375-382
• /
• 1999

Linear-by-linear association models are commonly used to analyze ordered categorical data. To fit these models appropriate scores need to be chosen. In this paper we perform sensitivity analyses in two-way contingency tables to investigate the effect of scores on goodness-of-fits and on tests of significance. In addition we show that the best score which yields the best fit of data can be selected based on the sensitivity analysis results.

• International Journal of Control, Automation, and Systems
• /
• v.6 no.3
• /
• pp.306-319
• /
• 2008

Reaching fault tolerance in technological systems requires to detect malfunctions. This paper presents a diagnostic method that is robust with respect to unknown-but-bounded uncertainties of the dynamical model and the measurements. By using models of the faultless and the faulty behaviours, a state-set observer computes polyhedral sets from which the consistency of the models with the interval measurements is determined. The diagnostic result is proven to be complete, i.e., the set of faults obtained by the diagnostic algorithm includes the actual fault. The algorithm is illustrated by an application example.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.27 no.2
• /
• pp.1-13
• /
• 2002

In this paper, we compares the business failure prediction accuracy among Linear Programming Discriminant Analysis(LPDA) model, Multivariate Discriminant Analysis (MDA) model and logit analysis model. The Data for 417 companies analyzed were gathered from KIS-FAS Published by Korea Information Service in 1999. The result of comparison for four time horizons shows that LPDA Is advantageous in prediction accuracy over the other two models when over all tilt ratio and business failure accuracy are considered simultaneously.

• Journal of the Korean Statistical Society
• /
• v.21 no.1
• /
• pp.70-79
• /
• 1992

The generalized linear models with random regrssors case are studied for bootstrapping. Only the natural link functions are considered. It is shown that the bootstrap approximation to the distribution of the maximum likelihood estimators is valid for almost all sample sequences. A slight extension of this model is also considered.

• Communications for Statistical Applications and Methods
• /
• v.2 no.2
• /
• pp.310-319
• /
• 1995

We represent graphically the relationship of hierachical log linear models by regarding the values of the likelihood ratio statistics as the squared norm of the corresponding vectors. Right angled triangles, tetrahedrons, and modified polyhedrons are used for graphical description. We find that the angle between the two vectors depends on the coefficient of determination and the partial coefficent of determination. Thess graphical descriptions could be applied to the model selection method.