• The Korean Journal of Applied Statistics
• /
• v.8 no.2
• /
• pp.147-161
• /
• 1995

The primary objective of this paper is to review parametric models and test statistics related to overdspersion of count data. Poisson or binomial assumption often fails to explain overdispersion. We reviewed real examples of overdispersion in count data that occurred in toxicological or teratological experiments. We also reviewed several models that were suggested for implementing experiments. We also reviewed several models that were suggested for implementing the extra-binomial variation or hyper-Poisson variability, and we noted how these models were generalized and further developed. The approaches that have been suggested for the overdispersion fall into two broad categories. The one is to develop a parametric model for it, and the other is to assume a particular relationship between the variance and the mean of the response variable and to derive a score test staistics for detecting the overdispersion. Recently, Dean(1992) derived a general score test statistics for detecting overdispersion from the exponential family.

• The Korean Journal of Applied Statistics
• /
• v.16 no.2
• /
• pp.377-393
• /
• 2003

Several confidence interval estimates for the difference of two binomial proportions were introduced. Bootstrap confidence interval is also suggested. We examined the over estimation property of approximate intervals and under estimation trend of exact intervals for the difference of proportions. We compared these confidence intervals based on the average coverage probability, expected width and skewness measure. Particularly actual coverage probability were calculated by using the prior distribution of parameters. Monte Carlo simulation for small sample size is conducted. Some interesting contour plots of average coverage probability and marginal plots for several interval estimates are presented.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.14 no.1
• /
• pp.85-93
• /
• 2015

Previous studies have estimated crash prediction models with the fixed effect model which assumes the fixed value of coefficients without considering characteristics of each intersections. However the fixed effect model would estimate under estimation of the standard error resulted in over estimation of t-value. In order to overcome these shortcomings, the random effect model can be used with considering heterogeneity of AADT, geometric information and unobserved factors. In this study, data collections from 89 intersections in Daejeon and estimates of crash prediction models were conducted using the random and fixed effect negative binomial regression model for comparison and analysis of two models. As a result of model estimates, AADT, speed limits, number of lanes, exclusive right turn pockets and front traffic signal were found to be significant. For comparing statistical significance of two models, the random effect model could be better statistical significance with -1537.802 of log-likelihood at convergence comparing with -1691.327 for the fixed effect model. Also likelihood ration value was computed as 0.279 for the random effect model and 0.207 for the fixed effect model. This mean that the random effect model can be improved for statistical significance of models comparing with the fixed effect model.

• Journal of Korean Society of Transportation
• /
• v.32 no.4
• /
• pp.369-379
• /
• 2014

The Statistical regression model has been used to construct crash prediction models, despite its limitations in assuming data distribution and functional form. In response to the limitations associated with the statistical regression models, a few studies based on non-parametric methods such as neural networks have been proposed to develop crash prediction models. However, these models have a major limitation in that they work as black boxes, and therefore cannot be directly used to identify the relationships between crash frequency and crash factors. A genetic programming model can find a solution to a problem without any specified assumptions and remove the black box effect. Hence, this paper investigates the application of the genetic programming technique to develope the crash prediction model. The data collected from the Gyeongbu expressway during the past three years (2010-2012), were separated into straight and curve sections. The random forest technique was applied to select the important variables that affect crash occurrence. The genetic programming model was developed based on the variables that were selected by the random forest. To test the goodness of fit of the genetic programming model, the RMSE of each model was compared to that of the negative binomial regression model. The test results indicate that the goodness of fit of the genetic programming models is superior to that of the negative binomial models.

• Journal of the Korean Society for Railway
• /
• v.20 no.2
• /
• pp.265-274
• /
• 2017

For accurate demand forecasting of railway logistics, we estimated intercity freight mode choice models based on the binary logit model and using production-consumption data from the Korea Transport Database. We estimated two types of models and compared the results by major item of railway logistics, such as container, cement, and steel: 1) The aggregate freight mode choice models are based on the revealed preference (RP) data and 2) The disaggregate models are based on the stated preference (SP) data. With respect to the container, the travel time variable was found to be statistically significant; however, the travel cost variable was not statistically significant in the RP model, while the travel cost variable was statistically significant in the SP model. For cement and steel, the travel cost variables were statistically significant but the travel time variables were not statistically significant in either the RP or the SP models. These results are inconsistent with results from previous studies based on SP data, which showed that the travel time variables were significant. Consequently, it can be concluded that the travel time factor should be considered in container transport, but that this factor is negligible for cement and steel transport.

• Journal of the Korean Data and Information Science Society
• /
• v.25 no.6
• /
• pp.1507-1520
• /
• 2014

In this paper, we show that there is a positive correlation between the level and the volatility of interest rate and thus suggest that a proper interest rate volatility coefficient (IRVC), a factor used in evaluating the interest rate risk that insurers are exposed to, should be chosen in accordance with the level of interest rate. To this end, we calculate the historical volatility of interest rate using data on government bond yields and show a proportionate relationship between interest rate and historical volatility. The review of exponential Vasicek (EV) and Cox-Ingersoll-Ross (CIR) models for interest rate also confirms the positive correlation between them. The estimation of IRVC by EV and CIR models are 0.9 and 1.1, respectively, which are much smaller than the one under the current risk-based capital (RBC) requirement. We provide modified IRVCs reflecting the level of interest by the two interest rate models. Using modified IRVCs can be a more reasonable method to evaluate the interest rate risk that insurers face.

• Journal of Animal Science and Technology
• /
• v.46 no.2
• /
• pp.137-144
• /
• 2004

Algorithms for estimating breeding values on several categorical data by using latent variables with threshold conception were developed and showed. Thresholds on each categorical trait were estimated by Newton’s method via gradients and Hessian matrix. This algorithm was developed by way of expansion of bivariate analysis provided by Quaas(2001). Breeding values on latent variables of categorical traits and observations on linear traits were estimated by preconditioned conjugate gradient(PCG) method, which was known having a property of fast convergence. Example was shown by simulated data with two linear traits and a categorical trait with four categories(CE=calving ease) and a dichotomous trait(SB=Still Birth) in threshold animal mixed model(TAMM). Breeding value estimates in TAMM were compared to those in linear animal mixed model (LAMM). As results, correlation estimates of breeding values to parameters were 0.91${\sim}$0.92 on CE and 0.87${\sim}$0.89 on SB in TAMM and 0.72~0.84 on CE and 0.59~0.70 on SB in LAMM. As conclusion, PCG method for estimating breeding values on several categorical traits with linear traits were feasible in TAMM.

• The Korean Journal of Applied Statistics
• /
• v.26 no.2
• /
• pp.267-280
• /
• 2013

College tuition is a significant economic, social, and political issue in Korea. We conduct a Bayesian analysis of a hierarchical model to address the factors related to college tuition based on a survey data collected by Statistics Korea. A binary response variable is selected depending on if more than 70% of tuition costs are supported by parents, and a hierarchical Probit model is constructed with areas as groups. A set of explanatory variables is selected from a factor analysis of available variables in the survey. A Markov chain Monte Carlo algorithm is used to estimate parameters. From the analysis results, income and stress are significantly related to college tuition support from parents. Parents with high income tend to support children's college tuition and students with parents' financial support tend to be mentally less stressed; subsequently, this shows that the economic status of parents significantly affects the mental health of college students. Gender, a healthy life style, and college satisfaction are not significant factors. Comparing areas in terms of the degrees of correlation between stress/income and tuition support from parents, students in Kangwon-do are the most mentally stressed when parents' support is limited; in addition, the positive correlation between parents support and income is stronger in big cities compared to provincial areas.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.30 no.5D
• /
• pp.453-459
• /
• 2010

In the study of traffic safety, it is utmost important to obtain more reliable estimates of the expected crashes for a site (or a segment). The observed crashes have been mainly used as the estimate of the expected crashes in Korea, while the empirical Bayes (EB) estimates based on the Poisson-gamma mixture model have been used in the USA and several European countries. Although numerous studies have used the EB method for estimating the expected crashes and/or the effectiveness of the safety countermeasures, no past studies examine the difference in the estimation errors between the two estimates. Thus, this study compares the estimation errors of the two estimates using a Monte Carlo simulation study. By analyzing the crash dataset at 3,000,000 simulated sites, this study reveals that the estimation errors of the EB estimates are always less than those of the observed crashes. Hence, it is imperative to incorporate the EB method into the traffic safety research guideline in Korea. However, the results show that the differences in the estimation errors between the two estimates decrease as the uncertainty of the prior distribution increases. Consequently, it is recommended that the EB method be used with reliable hyper-parameter estimates after conducting a comprehensive examination on the estimated negative binomial model.