• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.33 no.11
• /
• pp.1314-1319
• /
• 2009

The design methods of mechanical systems are largely classified into deterministic methods and stochastic methods. In deterministic methods, design parameters are assumed to have fixed values. On the other hand, in stochastic methods, design parameters are assumed to be statistically distributed. When a stochastic method is employed, statistical characteristics of the populations of design variables are assumed to be known. However, very often, it is almost impossible or very expensive to obtain the statistical characteristics of the populations. Therefore a sample survey method is usually employed for stochastic methods. This paper describes the procedure of estimating the statistical characteristics of populations by employing sample data sets. An example of AFM micro cantilever beam is employed to show the effectiveness of the procedure.

• Communications for Statistical Applications and Methods
• /
• v.18 no.1
• /
• pp.147-153
• /
• 2011

In this paper a new sampling procedure for estimating the population mean is introduced. The performance of the new population mean estimator is discussed, along with its properties, and it is shown that the proposed method generates an unbiased estimator. The relative efficiency of the suggested estimator is computed, in regards to the simple random sample(SRS), and comparisons are made to the ranked set sampling(RSS) and extreme ranked set sampling(ERSS) estimators used for asymmetric distributions. The results indicate that the proposed estimator is more efficient than the estimators based on the ERSS. In addition, the folded ranked set sampling(FRSS) procedure has an advantage over the RSS and ERSS in that it reduces the number of unused sampling units.

• Proceedings of the Korea Society for Simulation Conference
• /
• 1998.03a
• /
• pp.101-105
• /
• 1998

A large system predictor, which can perform prediction of sales trend in a huge number of distribution centers, is presented using neural predictive model. There are 20,000 number of distribution centers, and each distribution center need to forecast future demand in order to establish a reasonable inventory policy. Therefore, the number of forecasting models corresponds to the number of distribution centers, which is not possible to estimate that kind of huge number of accurate models in ERP (Enterprise Resource Planning)module. Multilayer neural net as universal approximation is employed for fitting the prediction model. In order to improve prediction accuracy, a sequential simulation procedure is performed to get appropriate network structure and also to improve forecasting accuracy. The proposed simulation procedure includes neural structure identification and virtual predictive model generation. The predictive model generation consists of generating virtual signals and estimating predictive model. The virtual predictive model plays a key role in tuning the real model by absorbing the real model errors. The complement approach, based on real and virtual model, could forecast the future demands of various distribution centers.

• Communications for Statistical Applications and Methods
• /
• v.24 no.6
• /
• pp.651-662
• /
• 2017

We propose an alternative procedure to select penalty parameter in $L_1$ penalized robust regression. This procedure is based on marginalization of prior distribution over the penalty parameter. Thus, resulting objective function does not include the penalty parameter due to marginalizing it out. In addition, its estimating algorithm automatically chooses a penalty parameter using the previous estimate of regression coefficients. The proposed approach bypasses cross validation as well as saves computing time. Variable-wise penalization also performs best in prediction and variable selection perspectives. Numerical studies using simulation data demonstrate the performance of our proposals. The proposed methods are applied to Boston housing data. Through simulation study and real data application we demonstrate that our proposals are competitive to or much better than cross-validation in prediction, variable selection, and computing time perspectives.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.33B no.7
• /
• pp.183-188
• /
• 1996

In this paper, we propose a new method which estimates the initial position of a PM synchronous motor (PMSM) using an incremental ecoder. in cremental encoders are widely used in detecting rotor position since they are inexpensive. At starting procedure, however, the incremental encoders cannot give the information of the absolute rotor position. Hence, additional absolue encoder, which has $60^{\circ}C$ resolutio electrically, and signal lines called U, V, W, have been used in practical systems for starting injection of a high frequency test current. It does not require any additional information such as U, V, W signals for starting procedure and allows for cheap and simple implementation. Experimental results show that the proposed method can estimate the initial rotor positon with higher accuracy than the conventional method using U, V, W signals.

• Journal of the Korean Data and Information Science Society
• /
• v.23 no.4
• /
• pp.825-832
• /
• 2012

We propose a novel way of forecasting the market shares of several brands simultaneously in a multiplicative competitive interaction model, which uses kernel regression technique incorporated with kernel machine technique applied in support vector machines and other machine learning techniques. Traditionally, the estimations of the market share attraction model are performed via a maximum likelihood estimation procedure under the assumption that the data are drawn from a normal distribution. The proposed method is shown to be a good candidate for forecasting method of the market share attraction model when normal distribution is not assumed. We apply the proposed method to forecast the market shares of 4 Korean car brands simultaneously and represent better performances than maximum likelihood estimation procedure.

• Communications for Statistical Applications and Methods
• /
• v.12 no.2
• /
• pp.305-312
• /
• 2005

In the cluster analysis, Hotelling's $T^2$ can be used to estimate the unknown number of clusters based on the idea of multiple comparison procedure. Especially, its threshold is obtained according to the probability of committing the type one error. Examples are used to compare Hotelling's $T^2$ with other classical location test statistics such as Sum-of-Squared Error and Wilks' $\Lambda$ The hierarchical clustering is used to reveal the underlying structure of the data. Also related criteria are reviewed in view of both the between variance and the within variance.

• Communications for Statistical Applications and Methods
• /
• v.6 no.3
• /
• pp.957-966
• /
• 1999

We evaluate the accuacy and precision of maximum likelihood estimation procedures for infectivity of HIV in partner studies. This is achieved by applying the oricedyre typothetical samples generated by computer. One hundred samples were generated with various combinations of parameters. The estimation procedure was found to be quite accurate. in addition it was found that the power of the test for equality of infectivities for two types of contact depends on sample size and length of observation period but not on the number of observations made on each subject. Tests based on a model for the infectivity had higher power than standard methods for comparing proportions.

• Proceedings of the Earthquake Engineering Society of Korea Conference
• /
• 2003.03a
• /
• pp.12-15
• /
• 2003

Korea Institute of Geoscience and Mineral Resources(KIGAM) is operating Wonju Korea Seismic Research Station(KSRS), 29 regional seismic research stations and 5 Korea-China joint seismic stations in China. Also KIGAM is operating Korea Earthquake Monitoring System (KEMS) to archive the real-time data stream and to determine event parameters (epicenter, origin time, and magnitude) by the automatic processing and analyst review. To do this, KEMS used KIGAM's regional seismic network and other institute's network in a near real-time base. From Dec. 1, 2001 to Nov. 30, 2002, 3,827 seismic events were analyzed in a automatic processing procedure and finally 3,437 events were analyzed by analyst and archived. But problem is this event catalog includes not only natural earthquake, but also artificial events produced by the blast. More than 80 % events were concentrated in daytime and many events were concentrated in the confirmed blast sites, Pyeongyang, Pocheon, Yeongjong-do, Donghae city, etc. Because these artificial events are a major potential cause of error when estimating the seismicity of a specific region, discrimination procedure has to be developed in the first place.

• Korean Journal of Hydrosciences
• /
• v.6
• /
• pp.13-22
• /
• 1995

The augmenting hydrologic data using a correlation procedure has been used to improve the estimates of the mean and variance at the site of interest with short record when one or more near by sites with longer records are available. The variance of the unbiased maximum likelihood estimator of $ derived by Moran based on the multivariate normal distribytion is modified into the form of Matalas and Jacobs for the biveriate normal distribution to get the critical minimum values of the multiple correlation coefficient which give the improvement for estimating the variance at the site of interest. Those values are tabulated for various lengths of short records and the number of sites.