• The Korean Journal of Applied Statistics
• /
• v.34 no.5
• /
• pp.697-710
• /
• 2021

Relative error prediction is preferred over ordinary prediction methods when relative/percentile errors are regarded as important, especially in econometrics, software engineering and government official statistics. The relative error prediction techniques have been developed in linear/nonlinear regression, nonparametric regression using kernel regression smoother, and stationary time series models. However, random effect models have not been used in relative error prediction. The purpose of this article is to extend relative error prediction to some of generalized linear mixed model (GLMM) with panel data, which is the random effect models based on gamma, lognormal, or inverse gaussian distribution. For better understanding, the real auto insurance data is used to predict the claim size, and the best predictor and the best relative error predictor are comparatively illustrated.

• Animal Bioscience
• /
• v.34 no.10
• /
• pp.1623-1631
• /
• 2021

Objective: This study aimed to validate and evaluate the dry matter (DM) intake prediction model of the Korean feeding standards for dairy cattle (KFSD). Methods: The KFSD DM intake (DMI) model was developed using a database containing the data from the Journal of Dairy Science from 2006 to 2011 (1,065 observations 287 studies). The development (458 observations from 103 studies) and evaluation databases (168 observations from 74 studies) were constructed from the database. The body weight (kg; BW), metabolic BW (BW^0.75, MBW), 4% fat-corrected milk (FCM), forage as a percentage of dietary DM, and the dietary content of nutrients (% DM) were chosen as possible explanatory variables. A random coefficient model with the study as a random variable and a linear model without the random effect was used to select model variables and estimate parameters, respectively, during the model development. The best-fit equation was compared to published equations, and sensitivity analysis of the prediction equation was conducted. The KFSD model was also evaluated using in vivo feeding trial data. Results: The KFSD DMI equation is 4.103 (±2.994)+0.112 (±0.022)×MBW+0.284 (±0.020)×FCM-0.119 (±0.028)×neutral detergent fiber (NDF), explaining 47% of the variation in the evaluation dataset with no mean nor slope bias (p>0.05). The root mean square prediction error was 2.70 kg/d, best among the tested equations. The sensitivity analysis showed that the model is the most sensitive to FCM, followed by MBW and NDF. With the in vivo data, the KFSD equation showed slightly higher precision (R² = 0.39) than the NRC equation (R² = 0.37), with a mean bias of 1.19 kg and no slope bias (p>0.05). Conclusion: The KFSD DMI model is suitable for predicting the DMI of lactating dairy cows in practical situations in Korea.

• Nuclear Engineering and Technology
• /
• v.56 no.5
• /
• pp.1747-1753
• /
• 2024

The accurate determination of formation density and the physical properties of rocks is the most critical logging tasks which can be obtained using gamma-ray transport and detection tools. Though the simulation works published so far have considerably improved the knowledge of the parameters that govern the responses of the detectors in these tools, recent studies have found considerable differences between the results of using a conventional model of a homogeneous mixture of formation and fluid and an inhomogeneous fractured medium. It has increased concerns about the importance of the complexity of the model used for the medium in simulation works. In the present study, we have suggested two various models for the flow of the fluid in porous media and fractured rock to be used for logging purposes. For a typical gamma-gamma logging tool containing a ¹³⁷Cs source and two NaI detectors, simulated by using the MCNPX code, a simplified porous (SP) model in which the formation is filled with elongated rectangular cubes loaded with either mineral material or oil was investigated. In this model, the oil directly reaches the top of the medium and the connection between the pores is not guaranteed. In the other model, the medium is a large 3-D matrix of 1 cm³ randomly filled cubes. The designed algorithm to fill the matrix sites is so that this realistic random (RR) model provides the continuum growth of oil flow in various disordered directions and, therefore, fulfills the concerns about modeling the rock textures consist of extremely complex pore structures. For an arbitrary set of oil concentrations and various formation materials, the response of the detectors in the logging tool has been considered as a criterion to assess the effect of modeling for the distribution of pores in the formation on simulation studies. The results show that defining a RR model for describing heterogeneities of a porous medium does not effectively improve the prediction of the responses of logging tools. Taking into account the computational cost of the particle transport in the complex geometries in the Monte Carlo method, the SP model can be satisfactory for gamma-gamma logging purposes.

• Magazine of the Korean Society of Agricultural Engineers
• /
• v.22 no.3
• /
• pp.75-87
• /
• 1980

Most hydro]ogic phenomena are the complex and organic products of multiple causations like climatic and hydro-geological factors. A certain significant correlation on the run-off in river basin would be expected and foreseen in advance, and the effect of each these causual and associated factors (independant variables; present-month rainfall, previous-month run-off, evapotranspiration and relative humidity etc.) upon present-month run-off(dependent variable) may be determined by multiple regression analysis. Functions between independant and dependant variables should be treated repeatedly until satisfactory and optimal combination of independant variables can be obtained. Reliability of the estimated function should be tested according to the result of statistical criterion such as analysis of variance, coefficient of determination and significance-test of regression coefficients before first estimated multiple regression model in historical sequence is determined. But some error between observed and estimated run-off is still there. The error arises because the model used is an inadequate description of the system and because the data constituting the record represent only a sample from a population of monthly discharge observation, so that estimates of model parameter will be subject to sampling errors. Since this error which is a deviation from multiple regression plane cannot be explained by first estimated multiple regression equation, it can be considered as a random error governed by law of chance in nature. This unexplained variance by multiple regression equation can be solved by stochastic approach, that is, random error can be stochastically simulated by multiplying random normal variate to standard error of estimate. Finally hybrid model on estimation of monthly run-off in nonhistorical sequence can be determined by combining the determistic component of multiple regression equation and the stochastic component of random errors. Monthly run-off in Naju station in Yong-San river basin is estimated by multiple regression model and hybrid model. And some comparisons between observed and estimated run-off and between multiple regression model and already-existing estimation methods such as Gajiyama formula, tank model and Thomas-Fiering model are done. The results are as follows. (1) The optimal function to estimate monthly run-off in historical sequence is multiple linear regression equation in overall-month unit, that is; Qn=0.788Pn+0.130Qn-1-0.273En-0.1 About 85% of total variance of monthly runoff can be explained by multiple linear regression equation and its coefficient of determination (R2) is 0.843. This means we can estimate monthly runoff in historical sequence highly significantly with short data of observation by above mentioned equation. (2) The optimal function to estimate monthly runoff in nonhistorical sequence is hybrid model combined with multiple linear regression equation in overall-month unit and stochastic component, that is; Qn=0. 788Pn+0. l30Qn-1-0. 273En-0. 10+Sy.t The rest 15% of unexplained variance of monthly runoff can be explained by addition of stochastic process and a bit more reliable results of statistical characteristics of monthly runoff in non-historical sequence are derived. This estimated monthly runoff in non-historical sequence shows up the extraordinary value (maximum, minimum value) which is not appeared in the observed runoff as a random component. (3) "Frequency best fit coefficient" (R2f) of multiple linear regression equation is 0.847 which is the same value as Gaijyama's one. This implies that multiple linear regression equation and Gajiyama formula are theoretically rather reasonable functions.

• Journal of Information Technology Services
• /
• v.17 no.1
• /
• pp.91-103
• /
• 2018

This study developed a prediction model using machine learning technology and predicted the success of health consulting by using life log data generated through u-Health service. The model index of the Random Forest model was the highest using. As a result of analyzing the Random Forest model, blood pressure was the most influential factor in the success or failure of metabolic syndrome in the subjects of u-Health service, followed by triglycerides, body weight, blood sugar, high cholesterol, and medication appear. muscular, basal metabolic rate and high-density lipoprotein cholesterol were increased; waist circumference, Blood sugar and triglyceride were decreased. Further, biometrics and health behavior improved. After nine months of u-health services, the number of subjects with four or more factors for metabolic syndrome decreased by 28.6%; 3.7% of regular drinkers stopped drinking; 23.2% of subjects who rarely exercised began to exercise twice a week or more; and 20.0% of smokers stopped smoking. If the predictive model developed in this study is linked with CBR, it can be used as case study data of CBR with high probability of success in the prediction model to improve the compliance of the subject and to improve the qualitative effect of counseling for the improvement of the metabolic syndrome.

• The Korean Journal of Applied Statistics
• /
• v.23 no.3
• /
• pp.511-520
• /
• 2010

The Pharmacokinetic model is a complex nonlinear model with pharmacokinetic parameters that is some-times represented by a complex form of differential equations. A population pharmacokinetic model adds individual variability using the random effects to the pharmacokinetic model. It amounts to the nonlinear mixed effect model. This paper, reviews the population pharmacokinetic model from a statistical viewpoint; in addition, a population pharmacokinetic model is also applied to the real clinical data along with a review of the statistical meaning of this model.

• Communications for Statistical Applications and Methods
• /
• v.12 no.3
• /
• pp.759-771
• /
• 2005

Twin studies are one of the most widely used methods for quantifying the influence of genetic and environmental factors on some traits such as a life span or a disease. In this paper we propose a genetic mixed linear model for twin survival time data, which allows us to separate the genetic component from the environmental component. Inferences are based upon the hierarchical likelihood (h-likelihood), which provides a statistically efficient and simple unified framework for various random-effect models. We also propose a simple and fast computation method for analyzing a large data set on twin survival study. The new method is illustrated to the survival data in Swedish Twin Registry. A simulation study is carried out to evaluate the performance.

• Korean Journal of Materials Research
• /
• v.34 no.3
• /
• pp.152-162
• /
• 2024

In this study, we introduce a novel TiN/Ag embedded TiO₂/FTO resistive random-access memory (RRAM) device. This distinctive device was fabricated using an environmentally sustainable, solution-based thin film manufacturing process. Utilizing the peroxo titanium complex (PTC) method, we successfully incorporated Ag precursors into the device architecture, markedly enhancing its performance. This innovative approach effectively mitigates the random filament formation typically observed in RRAM devices, and leverages the seed effect to guide filament growth. As a result, the device demonstrates switching behavior at substantially reduced voltage and current levels, heralding a new era of low-power RRAM operation. The changes occurring within the insulator depending on Ag contents were confirmed by X-ray photoelectron spectroscopy (XPS) analysis. Additionally, we confirmed the correlation between Ag and oxygen vacancies (V_o). The current-voltage (I-V) curves obtained suggest that as the Ag content increases there is a change in the operating mechanism, from the space charge limited conduction (SCLC) model to ionic conduction mechanism. We propose a new filament model based on changes in filament configuration and the change in conduction mechanisms. Further, we propose a novel filament model that encapsulates this shift in conduction behavior. This model illustrates how introducing Ag alters the filament configuration within the device, leading to a more efficient and controlled resistive switching process.

• The Korean Journal of Applied Statistics
• /
• v.30 no.6
• /
• pp.917-929
• /
• 2017

Korean income data obtained from Korea Labor Panel Survey shows excessive zeros, which may not be properly explained by the Tobit model. In this paper, we analyze the data using a zero-inflated Tobit model to incorporate excessive zeros. A zero-inflated Tobit model consists of two stages. In the first stage, individuals with 0 income are divided into two groups: genuine zero group and random zero group. Individuals in the genuine zero group did not participate labor market since they have no intention to do so. Individuals in the random zero group participated labor market but their incomes are very low and truncated at 0. In the second stage, the Tobit model is assumed to a subset of data combining random zeros and positive observations. Regression models are employed in both stages to obtain the effect of explanatory variables on the participation of labor market and the income amount. Markov chain Monte Carlo methods are applied for the Bayesian analysis of the data. The proposed zero-inflated Tobit model outperforms the Tobit model in model fit and prediction of zero frequency. The analysis results show strong evidence that the probability of participating in the labor market increases with age, decreases with education, and women tend to have stronger intentions on participating in the labor market than men. There also exists moderate evidence that the probability of participating in the labor market decreases with socio-economic status and reserved wage. However, the amount of monthly wage increases with age and education, and it is larger for married than unmarried and for men than women.

• Water Engineering Research
• /
• v.3 no.1
• /
• pp.31-44
• /
• 2002

The effect of diurnal cycle, intermittent visit of observation satellite, sensor installation, partial coverage of remote sensing, heterogeneity of soil properties and precipitation to the soil moisture estimation error were analyzed to present the global sampling strategy of soil moisture. Three models, the theoretical soil moisture model, WGR model proposed Waymire of at. (1984) to generate rainfall, and Turning Band Method to generate two dimensional soil porosity, active soil depth and loss coefficient field were used to construct sufficient two-dimensional soil moisture data based on different scenarios. The sampling error is dominated by sampling interval and design scheme. The effect of heterogeneity of soil properties and rainfall to sampling error is smaller than that of temporal gap and spatial gap. Selecting a small sampling interval can dramatically reduce the sampling error generated by other factors such as heterogeneity of rainfall, soil properties, topography, and climatic conditions. If the annual mean of coverage portion is about 90%, the effect of partial coverage to sampling error can be disregarded. The water retention capacity of fields is very important in the sampling error. The smaller the water retention capacity of the field (small soil porosity and thin active soil depth), the greater the sampling error. These results indicate that the sampling error is very sensitive to water retention capacity. Block random installation gets more accurate data than random installation of soil moisture gages. The Walnut Gulch soil moisture data show that the diurnal variation of soil moisture causes sampling error between 1 and 4 % in daily estimation.