• Communications for Statistical Applications and Methods
• /
• v.10 no.2
• /
• pp.585-594
• /
• 2003

It has been generally recognized that conventional binomial or Poisson model provides poor fits to the actual correlated binary data due to the extra-binomial variation. A number of generalized statistical models have been proposed to account for this additional variation. Among them, beta-binomial, correlated-binomial, and modified-binomial models are binomial-related models which are frequently used in modeling the sum of n correlated binary data. In many situations, it is reasonable to assume that n correlated binary data are exchangeable, which is a special case of correlated binary data. The sum of n exchangeable correlated binary data is modeled relatively well when the above three binomial-related models are applied. But the estimation results of correlation coefficient turn to be quite different. Hence, it is important to identify which model provides better estimates of model parameters(success probability, correlation coefficient). For this purpose, a small-scale simulation study is performed to compare the behavior of above three models.

• 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.

• Journal of the Korean Statistical Society
• /
• v.32 no.4
• /
• pp.385-399
• /
• 2003

We first introduce the quasi-score estimating function and applied the quasi-score estimating function to nonlinear time series models. We proposed the M quasi-score estimating functions bounded functions for the quasi-score estimating functions. Also, we investigated the asymptotic properties of quasi-likelihood estimators and M quasi-likelihood estimators. Simulation results show that the M quasi-likelihood estimators work better than the least squares estimators under the heavy-tailed distributions

• Journal of Korea Water Resources Association
• /
• v.44 no.2
• /
• pp.135-144
• /
• 2011

Direct parameter estimation method is verified with various models based on Neyman-Scott rectangular pulse model (NSRPM). Also, newly modified NSRPM (NMSRPM) that uses normal distribution is developed. Precipitation data observed by Korea Meteorological Administration (KMA) for 47 years is applied for parameter estimation. For model performance verification, we used statistics, wet ratio and precipitation accumulate distribution of precipitation generated. The comparison of statistics indicates that absolute relative error (ARE)s of the results from NSRPM and modified NSRPM (MNSRPM) are increasing on July, August, and September and ARE of NMNSRPM shows 10.11% that is the smallest ARE among the three models. NMNSRPM simulates the characteristics of precipitation statistics well. By comparing the wet ratio, MNSRPM shows the smallest ARE that is 16.35% and by using the graphical analysis, we found that these three models underestimate the wet ratio. The three models show about 2% of ARE of precipitation accumulate probability. Those results show that the three models simulate precipitation accumulate probability well. As the results, it is found that the parameters of NSRPM, MNSRPM and NMNSRPM are able to be estimated by the direct parameter estimation method. From the results listed above, we concluded that the direct parameter estimation is able to be applied to various models based on NSRPM. NMNSRPM shows good performance compared with developed model-NSRPM and MNSRPM and the models based on NSRPM can be developed by the direct parameter estimation method.

• Journal of Information Technology Services
• /
• v.5 no.1
• /
• pp.97-109
• /
• 2006

As software becomes more complex and its scope dramatically increases, the importance of research on developing methods for estimating software development efforts has been increased. Such accurate estimation has a prominent impact on the development projects. To develop accurate effort estimation models, many studies have been conducted among the academia and the practitioners. Out of the numerous methods, Constructive Cost Model (COCOMO) based on Line of Code (LOC), Regression Model based on Function Point (FP) were the most popular models in the past. As today's development environments are dynamically changing, these traditional methods do not work anymore. There is an impending need to develop an accurate estimation model which accommodates itself to the new environments. As a possible solution, this research proposes and evaluates an software development estimation model based on function points and neural networks.

• The Korean Journal of Applied Statistics
• /
• v.26 no.1
• /
• pp.163-185
• /
• 2013

In this paper, we investigate the need to employ long-memory volatility models in terms of Value-at-Risk(VaR) estimation. We estimate the VaR of the KOSPI returns using long-memory volatility models such as FIGARCH and FIEGARCH; in addition, via back-testing we compare the performance of the obtained VaR with short memory processes such as GARCH and EGARCH. Back-testing says that there exists a long-memory property in the volatility process of KOSPI returns and that it is essential to employ long-memory volatility models for the right estimation of VaR.

• Honam Mathematical Journal
• /
• v.31 no.1
• /
• pp.109-124
• /
• 2009

Based on the observed data for Clarithromycin released, three commonly used inflow models: the power, the exponential, and the logarithmic models are considered. Among them, the power model is used most in practice for simplicity. Using the numerical parameter estimation techniques, the parameters appeared in the model equations are estimated. Through the numerical estimation results using the several experimental data sets, the exponential model turns out to be best among the three models. More specifically, the sum of squares of absolute errors and the sum of squares of relative errors for the exponential model are reduced by 80-95 % for the experimental data sets and 60-90 % for the noise added data sets compared with those for the power and logarithmic models. A typical experimental data set is used in this paper to show the estimation method and its numerical results. The proposed numerical method and its algorithm are designed for estimating the parameters appeared in the model differential equations for which the exact form of the solution is unknown in general. The methodology developed can be applied to more general cases such as the nonlinear ordinary differential equations or the partial differential equations.

• Journal of Korea Society of Digital Industry and Information Management
• /
• v.19 no.1
• /
• pp.109-118
• /
• 2023

Data-based analysis methods have become used more for estimating or predicting housing prices, and neural network models and decision trees in the field of big data are also widely used more and more. Neural network models are often evaluated to be superior to existing statistical models in terms of estimation or prediction accuracy. However, there is ambiguity in determining the input feature of the input layer of the neural network model, that is, the type and number of input features, and decision trees are sometimes used to overcome these disadvantages. In this paper, we evaluate the existing methods of using decision trees and propose the method of using decision trees to prioritize input feature selection in neural network models. This can be a complementary or combined analysis method of the neural network model and decision tree, and the validity was confirmed by applying the proposed method to house price estimation. Through several comparisons, it has been summarized that the selection of appropriate input characteristics according to priority can increase the estimation power of the model.

• Journal of the Korean Statistical Society
• /
• v.12 no.1
• /
• pp.24-35
• /
• 1983

In fitting models to data, selection of the most desirable estimation method and determination of the adequacy of fitted model are the central issues. This paper compares the maximum likelihood estimators and the minimum logit chi-square estimators, both being best asymptotically normal, when logit models are fitted to infant mortality data. Chi-square goodness-of-fit test and likelihood ratio one are also compared. The analysis infant mortality data shows that the outlying observations do not necessarily result in the same impact on goodness-of-fit measures.

• Proceedings of the Korean Statistical Society Conference
• /
• 2002.11a
• /
• pp.3-9
• /
• 2002

In financial time series, the autoregressive conditional heteroscedastic (ARCH) models have been widely used for modeling conditional variances. In many cases, non-normality or heavy-tailed distributions of the data have influenced the estimation methods under normality assumption. To solve this problem, a robust function for the conditional variances of the errors is proposed and compared the relative efficiencies of the estimators with other conventional models.