• Journal of Korean Society of Industrial and Systems Engineering
• /
• v.32 no.1
• /
• pp.11-19
• /
• 2009

A number of trend test methods, i.e., Military Handbook test and Laplace test etc., have been applied to investigate recurrent failures trend in repairable systems. Existing methods provide information about only existence of trend in the system. In this paper, we propose a new change-point test based on the Schwarz Information Criterion(SIC). The change-point approach is more informative than other trend test methods in that it provides the number of change-points and the location of change-points if it exists, as well as the existence of change-point for recurrent failures. The change-point test is applied to nine 300MW generating units operated in East China. We confirm that the change-point test has a potential for establishing optimal preventive maintenance policy by detecting change-point of failure rate.

• Communications for Statistical Applications and Methods
• /
• v.7 no.3
• /
• pp.789-798
• /
• 2000

The mean residual function is the expected remaining life of an item at age x. The problem of trend change in the mean residual life is great interest in the reliability and survival analysis. In this paper, we develop a family of test statistics for testing whether or not the mean residual life changes its trend. The asymptotic normality of the test statistics is established. Monte Carlo simulations are conducted to study the performance of our test statistics.

• The Korean Journal of Applied Statistics
• /
• v.36 no.6
• /
• pp.573-589
• /
• 2023

In a piecewise linear trend model, the change points coincide with the mean change points of the first differenced time series. Therefore, by detecting the mean change points of the first differenced time series, one can estimate the change points of the piecewise linear trend model. In this paper, based on this fact, a method is proposed for detecting change points of the piecewise linear trend model using the simple moving average of the first differenced time series rather than estimates of the slope or residuals. Our Monte Carlo simulation experiments show that the proposed method performs well in estimating the number of change points not only when the error terms in the piecewise linear trend model are independent but also when they are serially correlated.

• Journal of Applied Reliability
• /
• v.3 no.1
• /
• pp.83-92
• /
• 2003

We studied the trend change of failure rate function and uncertainty of residual life function in terms of location of their trend change points. It is shown that the trend change of uncertainty of residual life takes place before the failure rate changes its trend. Like DIFR(IDFR) does not necessary implies IDMRL(DIMRL), we find the fact that DIFR(IDFR) does not always imply IDURL(DIURL) under certain conditions, through the exponentiated-weibull distribution.

• Journal of Korean Society for Quality Management
• /
• v.30 no.3
• /
• pp.195-202
• /
• 2002

The problem of trend change in the mean residual life is great Interest in the reliability and survival analysis. In this paper, a new test statistic for testing whether or not the mean residual life changes its trend Is developed. It is assumed that neither the change point nor the proportion at which the trend change occurs is known. The asymptotic null distribution of test statistic is established and asymptotic critical values of the asymptotic null distribution is obtained. Monte Carlo simulation is used to compare the proposed test with previously known tests.

• Journal of Korean Society for Quality Management
• /
• v.30 no.4
• /
• pp.78-85
• /
• 2002

The problem of trend change in the mean residual life is great interest in the reliability and survival analysis. In this paper, a new test statistic for testing whether or not the mean residual life changes its trend is developed. It is assumed that neither the change point nor the proportion at which the trend change occurs is known. The asymptotic null distribution of test statistic is established and asymptotic critical values of the asymptotic null distribution is obtained. Monte Carlo simulation is used to compare the proposed test with previously known tests.

• Journal of the Korean Society of Marine Environment & Safety
• /
• v.26 no.3
• /
• pp.233-241
• /
• 2020

We examined long-term variations in sea surface temperature (SST) and annual amplitudes of SST around the Korean Peninsula. Two SST data sets with data periods of approximately 51 years and longer than 100 years, respectively, were obtained from the National Institute of Fisheries Science and Japan Meteorological Agency. SST of Korean waters clearly increased during last 51 years (1968-2018), which was 2.5 times higher than the global trend. This significant increasing trend was caused by the dominant increasing SST trend during winter. However, a negative and positive SST anomaly frequently appeared during winter and summer, respectively, in a recent decade. These features of seasonal SST variation have changed the annual amplitude of SST, and resulted in a drastically increasing trend after 2009. Using the longer SST data set, it was revealed that the decreasing SST trend in winter began in the 2000s and the increasing SST trend in summer bagan in the 1990s. During a recent decade, there was a distinctive SST increase in summer, whereas a clear decrease in winter. In summary, the annual amplitude of SST around the Korean Peninsula significantly changed from a decreasing trend to an increasing trend during a recent decade.

• Communications for Statistical Applications and Methods
• /
• v.30 no.2
• /
• pp.149-162
• /
• 2023

The ℓ₁-trend filtering method is one of the most widely used methods for extracting underlying trends from noisy observations. Contrary to the Hodrick-Prescott filtering, the ℓ₁-trend filtering gives piecewise linear trends. One of the advantages of the ℓ₁-trend filtering is that it can be used for identifying change points in piecewise linear trends. However, since the ℓ₁-trend filtering employs total variation as a penalty term, estimated piecewise linear trends tend to be biased. In this study, we demonstrate the biasedness of the ℓ₁-trend filtering in trend level estimation and propose a two-stage bias-reduction procedure. The newly suggested estimator is based on the estimated change points of the ℓ₁-trend filtering. Numerical examples illustrate that the proposed method yields less biased estimates for piecewise linear trends.

• Atmosphere
• /
• v.31 no.5
• /
• pp.553-562
• /
• 2021

The long-term trend of surface wind speed in Korea is estimated by correcting wind measurements at 29 KMA weather stations from 1985 to 2019 with physical and statistical homogenization. The anemometer height changes at each station are first adjusted by applying physical homogenization using the power-law wind profile. The statistical homogenization is then applied to the adjusted data. A standard normal homogeneity test (SNHT) is particularly utilized. Approximately 40% of inhomogeneities detected by the SNHT match with the sea-level-height change of each station, indicating that an SNHT is an effective technique for reconciling data inhomogeneity. The long-term trends are compared with homogenized data. Statistically significant negative trends are observed along the coast, while insignificant trends are dominant inland. The mean trend, averaged over all stations, is -0.03 ± 0.07 m s^-1 decade^-1. This insignificant trend is due to a trend change across 2001. A decreasing trend of -0.10 m s^-1 decade^-1 reverses to an increasing trend of 0.03 m s^-1 decade^-1 from 2001. This trend change is consistent with mid-latitude wind change in the Northern hemisphere, indicating that the long-term trend of surface wind speed in Korea is partly determined by large-scale atmospheric circulation.

• Journal of the Korean Society of Clothing and Textiles
• /
• v.42 no.4
• /
• pp.708-725
• /
• 2018

The advancement of digital technology has made changes in the fashion system and trend development process inevitable. This article clarifies changes in the modern fashion industry system and the causes of comprehensive changes that result from the development of digital technology. The methodology of this study is based on literature and case studies based on the information magazine most used by fashion industry workers. This study classifies fashion systems into 5 types and 14 types in detail. The study results indicate the way to change the fashion style trend schedule per year, fast/ultrafast fashion system, fashion rental system, DTC system and change of fashion system by consumer participation. The causes of the changes in fashion system are indicated that an increase of trend sensitivity due to an increase in the diffusion rate of information, expansion of expression of personality through digital network, increase of possibility of grouping of small number of tastes and change of prosumer possible changes in the environment. This study provides basic data on fashion system research and the construction of an appropriate response strategy for a changing environment.