• The Korean Journal of Applied Statistics
• /
• v.29 no.4
• /
• pp.699-706
• /
• 2016

Outlier detection methods without performing a test often do not succeed in detecting multiple outliers because they are structurally vulnerable to a masking effect or a swamping effect. This paper considers testing procedures supplemented to a clustering-based method of identifying the group with a minority of the observations as outliers. One of general steps is performing a variety of t-test on individual outlier-candidates. This paper proposes a sequential procedure for searching for outliers by changing cutoff values on a cluster tree and performing a test on a set of outlier-candidates. The proposed method is illustrated and compared to existing methods by an example and Monte Carlo studies.

• Magazine of the Korean Society of Agricultural Engineers
• /
• v.30 no.4
• /
• pp.62-70
• /
• 1988

This study was conducted to solve the problems for the unsuitable parameters and the uncertainty of design flood can be appeared by low outliers were inclined to the lower part from the trend of the balance of the data. Derivation of reasonable design flood was attempted finally by modification of low outliers with analysis of flood frequency by means of Log Pearson Type Ill distribution. Three subwatersheds were selected as studying basins with the annual maximum series including low outliers along Geum River basin. The results through this study were analyzed and summarized as follows. 1. Log Pearson Type In distribution was confirmed as a reasonable one by X$^2$ goodness of fit test at Gong Ju, Gyu Am, og Cheon watershed along Geum River basin. 2. Probable flood flows for each watershed were derivated by flood frequency curve with outliers. 3. Weighted skew coefficient for each watershed was calculated for the evaluation of freq- uency factor which is needed for the modification of low outlier. 4. It was confirrned that adjusted frequency curve has a lower tendency than that of deletion of low outlier in common at all watersheds. 5. Final probable flood flows were derivated by modification with evaluation of modified basic statistics for three watersheds. 6. In comparison with a frequency curve with modification and one with outlier, The former has a higher probable flood flow within three years of return periods than that of the latter, and vice versa over three years of return periods.

• Journal of the Korean Data and Information Science Society
• /
• v.26 no.6
• /
• pp.1565-1572
• /
• 2015

In this paper, we analyze the polio incidence data based on the Poisson autoregressive models, focusing particularly on change-point detection. Since the data include some strongly deviating observations, we employ the robust cumulative sum (CUSUM) test proposed by Kang and Song (2015) to perform the test for parameter change. Contrary to the result of Kang and Lee (2014), our data analysis indicates that there is no significant change in the case of the CUSUM test with strong robustness and the same result is obtained after ridding the polio data of outliers. We additionally consider the comparison of the forecasting performance. All the results demonstrate that the robust CUSUM test performs adequately in the presence of seemingly outliers.

• The Journal of the Korean dental association
• /
• v.55 no.9
• /
• pp.604-616
• /
• 2017

In clinical dental research, errors occur in spite of careful study design and conduct. Data cleaning procedures intend to identify and correct these errors or at least to minimize their influence on study. Outlier is the one of these errors. Outlier detection is the first step in data analysis process which has a serious effect in the field of dental research. Hence, this paper aims to introduce the methods to detect the outliers and to examine their influences in statistical data analysis.

• Journal of Applied Reliability
• /
• v.17 no.3
• /
• pp.264-271
• /
• 2017

Purpose: Field operation data contains missing values or outliers due to various causes of the data collection process, so caution is required when utilizing RAM analysis results by field operation data. The purpose of this study is to present a method to minimize the RAM analysis error of the field data to improve the accuracy. Methods: Statistical methods are presented for processing of the outliers and the missing values of the field operating data, and after analyzing the RAM, the differences between before and after applying the technique are discussed. Results: The availability is estimated to be lower by 6.8 to 23.5% than that before processing, and it is judged that the processing of the missing values and outliers greatly affect the RAM analysis result. Conclusion: RAM analysis of OO weapon system was performed and suggestions for improvement of RAM analysis were presented through comparison with the new and current method. Data analysis results without appropriate treatment of error values may result in incorrect conclusions leading to inappropriate decisions and actions.

• The Korean Journal of Applied Statistics
• /
• v.27 no.2
• /
• pp.307-315
• /
• 2014

An exact distribution of the test statistic to test for multiple outlier candidates does not generally exist; therefore, tests of individual outliers (or tests using simulated critical-values) are usually conducted instead of testing for groups of outliers. This article is on procedures to test outlying observations. We suggest a method that can be applied to arbitrary observations or multiple outlier candidates detected by an outlier detecting method. A Monte Carlo study performance is used to compare the proposed method with others.

• Communications for Statistical Applications and Methods
• /
• v.7 no.2
• /
• pp.447-456
• /
• 2000

In this paper, we considered three methods for outlier identification sample surveys. First, we studied method of handling and adjusting outliers in normal population. Second, we studied existing methods using mean, maximum and minimum and proposed a test using of median which well reflects characteristic of data regardless of sampling distribution. Finally, we showed our test using median works better than Dixon and mean test through simulation.

• Proceedings of the Korean Statistical Society Conference
• /
• 2003.05a
• /
• pp.287-293
• /
• 2003

In this paper we consider the problem of parameter change based on the cusum test proposed by Lee et al. (2003). The cusum test statistic is constructed utilizing the estimator minimizing density-based divergence measures. It is shown that under regularity conditions, the test statistic has the limiting distribution of the sup of standard Brownian bridge. Simulation results demonstrate that the cusum test is robust when there arc outliers.

• Communications for Statistical Applications and Methods
• /
• v.4 no.2
• /
• pp.565-579
• /
• 1997

A robust test is proposed for the problem of testing the parallelism of several regression lines against ordered alternatives. The proposed test statistic is based on a linear combination of one-step pairwise GM-estimators. We compare the performance of the proposed test with that of the other tests through a Monte Carlo simulation. The results of the simulation study show that the proposed test has stable levels, good empirical powers in various circumstances, and particularly higher empirical powers under the presence of extreme outliers or leverage points.

• Journal of the Korean Data and Information Science Society
• /
• v.10 no.2
• /
• pp.369-385
• /
• 1999

For the growth curve model with arbitrary covariance structure, known as unstructured covariance matrix, the problems of detecting outliers are discussed in this paper. In order to detect outliers in the growth curve model, the likelihood ratio testing statistics in mean shift model is established and its distribution is derived. After we detected outliers in growth curve model, we test homo and/or hetero-geneous covariance matrices using PSR Quasi-Bayes Criterion. For illustration, one numerical example is discussed, which compares between before and after outlier deleting.