• Journal of the Korean Data and Information Science Society
• /
• v.24 no.2
• /
• pp.211-222
• /
• 2013

We consider the problem of nonparametrically estimating the conditional quantile function from censored data and propose new estimators here. They are based on local logistic regression technique of Lee et al. (2006) and "double-kernel" technique of Yu and Jones (1998) respectively, which are modified versions under random censoring. We compare those with two existing estimators based on a local linear fits using the check function approach. The comparison is done by a simulation study.

• The Korean Journal of Applied Statistics
• /
• v.29 no.5
• /
• pp.823-833
• /
• 2016

The lifetime is a key characteristic of product quality. It is best to obtain the lifetime data of all samples, but they are often censored due to time or expense limitations. In this paper, we propose a binomial cumulative sum (CUSUM) chart to monitor the mean of type I right-censored Weibull lifetime data, for a xed value of the Weibull shape parameter. We compare the performance of the proposed binomial CUSUM chart with CUSUM charts studied previously using the steady-state average run length (ARL). The results show that the performance of the binomial CUSUM chart is better when the censoring rate is high and/or the sample size is small.

• The Korean Journal of Applied Statistics
• /
• v.27 no.1
• /
• pp.1-11
• /
• 2014

The Weighted Logrank test and its special case, Logrank test are widely used to compare survival distributions; however, these methods are inappropriate when the sample size is small or censoring distributions are not equal since they use test statistics from approximate distributions. A permutation test can be an alternative for small sample cases; however, this should be used only when censoring distributions are equal. To handle cases with small sample size and unequal censoring distributions, the permutation-imputation method was developed to compare two survival distributions. In this paper, approximate method, permutation method and permutation-imputation method were compared using a Logrank test and Prentice-Wilcoxon test for three or more survival distributions comparison.

• Journal of the Korean Data and Information Science Society
• /
• v.23 no.4
• /
• pp.739-748
• /
• 2012

In this paper we propose a mixed effects least squares support vector machine (LS-SVM) for the censored data which are observed from different groups. We use weights by which the randomly right censoring is taken into account in the nonlinear regression. The weights are formed with Kaplan-Meier estimates of censoring distribution. In the proposed model a random effects term representing inter-group variation is included. Furthermore generalized cross validation function is proposed for the selection of the optimal values of hyper-parameters. Experimental results are then presented which indicate the performance of the proposed LS-SVM by comparing with a standard LS-SVM for the censored data.

• The Korean Journal of Applied Statistics
• /
• v.32 no.1
• /
• pp.129-138
• /
• 2019

Ammunition is an integral element of a weapon systems and in calculating fighting strength. The Korea Army utilizes the basic load (B/L) concept to supply ammunition smoothly. The required supply rate (RSR) is the basis of a B/L that is estimated from real combat data that includes a troop's mission and operation terrain. The current RSR is based on Korean War data and the sample mean has some problems in applications to modern combat. Therefore, this study used Korea Combat Training Center (KCTC) data that is similar to real combat to estimate rifle ammunition RSR. We used a quantile of truncated Weibull distribution to estimate rifle ammunition RSR considering that rifle ammunition consumption data in KCTC is truncated. As a result, we obtained a rifle ammunition RSR which covers most ammunition consumption by reflecting the individual consumption of rifle ammunition.

• Journal of the Korean Data and Information Science Society
• /
• v.22 no.2
• /
• pp.171-177
• /
• 2011

With the development of the actuarial and insurance industries, the distributions of the insurance payments data are deeply studied by many authors. It is known that theses types of distribution are very highly positively skewed and have a long thick upper tail such as Pareto or lognormal distribution. In 2005, Cooray and Ananda proposed a new model which is composed lognormal distribution and Pareto distribution. They said it as composite lognormal-Preto distribution. They showed that the proposed distribution was better fitted than lognormal or Pareto distribution. On the other hand many agreements about the insurance payment have some options for a trivially small payment or extremely large one because of the limits of total payment. Appling these cases, in this paper we consider the parameter estimation on the composite lognormal-Pareto distribution based on doubly censored samples.

• The Korean Journal of Applied Statistics
• /
• v.11 no.1
• /
• pp.113-127
• /
• 1998

There have been a great deal of interests in comparing two survival distributins in clinical trials. This paper compares some well-known statistical methods for testing the equality of two survival distributions. Simulation studies also provide some insights into the properties of these test statistics across several types of survival distributions and degrees of censorship.

• Journal of the Korean Society for Railway
• /
• v.16 no.3
• /
• pp.196-201
• /
• 2013

As there is no failure data for the entire lifecycle of a product, when analyzing reliability measures based on early failure data only, there may be a significant error between the estimated mean life and the real one, because it can be underestimated, or on the other hand, it can be overestimated when analyzing reliability measures based on a large amount of censored data with the failure data. To resolve the issue, this study proposes an optimal sampling estimation procedure that selects the proportion of censored data to estimate the optimal distribution with the idea that the estimated distribution could be approximated as closely as the real life distribution. This would work if we sampled the optimal proportion on the censored data, because failure data has real intrinsic distribution in any situation. We validate the proposed procedure using an actual example. If the proposed method is applied to the maintenance policy of TWC (Train to Wayside Communication) system, then we can establish the optimal maintenance policy. Thus, we expect that it will be effective for improvement of reliability and cost savings.

• The Korean Journal of Applied Statistics
• /
• v.25 no.2
• /
• pp.311-319
• /
• 2012

The simplest and the most important distribution in survival analysis is exponential distribution. Koziol and Green (1976) derived Cram$\acute{e}$r-von Mises statistic's randomly censored version based on the Kaplan-Meier product limit estimate of the distribution function; however, it could not be practical for a real data set since the statistic is for testing a simple goodness of fit hypothesis. We generalized it to the composite hypothesis for exponentiality with an unknown scale parameter. We also considered the classical Kolmogorov-Smirnov statistic and generalized it by the exact same way. The two statistics are compared through a simulation study. As a result, we can see that the generalized Koziol-Green statistic has better power in most of the alternative distributions considered.

• The Korean Journal of Applied Statistics
• /
• v.25 no.1
• /
• pp.183-196
• /
• 2012

The inference for incomplete data such as missing data, truncated distribution and censored data is a phenomenon that occurs frequently in statistics. To solve this problem, Expectation Maximization(EM), Monte Carlo Expectation Maximization(MCEM) and Stochastic Expectation Maximization(SEM) algorithm have been used for a long time; however, they generally assume known distributions. In this paper, we propose the Metropolis-Hastings Expectation Maximization(MHEM) algorithm for unknown distributions. The performance of our proposed algorithm has been investigated on simulated and real dataset, KOSPI 200.