• Journal of the Korean Data and Information Science Society
• /
• 제24권3호
• /
• pp.659-667
• /
• 2013

If the regression model or the propensity model is correct, the unbiasedness of the estimator using doubly robust imputation can be guaranteed. Using a neural network instead of a logistic regression model for the propensity model, the estimators using doubly robust imputation are approximately unbiased even though both assumed models fail. We also propose a doubly robust estimator of ratio form using population information of an auxiliary variable. We prove some properties of proposed theory by restricted simulations.

• 응용통계연구
• /
• 제31권4호
• /
• pp.463-473
• /
• 2018

연구에서는 선형회귀모형을 가정한 대형 데이터에서의 변수선택 알고리즘을 다룬다. 방법의 속도와 강건성에 주안점을 둔 여러 알고리즘들이 제안되었다. 그 중에서 streamwise 회귀 접근법을 사용한 VIF회귀는 신속하고 정확하게 수행된다. 그러나 VIF회귀는 최소제곱방법에 의해 모형이 추정되므로 이상치에 민감하다. 변수선택방법의 강건성을 높이기 위해 가중 추정치를 사용한 강건측도가 제안되었으며 강건 VIF회귀도 제안되었다. 본 연구에서는 잠재적 이상치를 탐지하여 제거한 후 VIF회귀를 수행하는, 빠르고 강건한 변수선택 방법을 제안한다. 제안된 방법은 모의실험과 데이터 분석 통해 다른 방법들과 비교된다.

• 산업공학
• /
• 제23권2호
• /
• pp.118-125
• /
• 2010

The determination of a regression model is important in using statistical designs of experiments. Generally, the exact regression model is not known, and experimenters suppose that a certain model form will be fit. Then an experimental design suitable for that predetermined model form is selected and the experiment is conducted. However, the initially chosen regression model may not be correct, and this can result in undesirable statistical properties. We develop model-robust experimental designs that have stable prediction variance for a family of candidate regression models over a cuboidal region by using genetic algorithms and the desirability function method. We then compare the stability of prediction variance of model-robust experimental designs with those of the 3-level face centered cube. These model-robust experimental designs have moderately high G-efficiencies for all candidate models that the experimenter may potentially wish to fit, and outperform the cuboidal design for the second-order model. The G-efficiencies are provided for the model-robust experimental designs and the face centered cube.

• Journal of the Korean Data and Information Science Society
• /
• 제15권2호
• /
• pp.307-316
• /
• 2004

Principal Component Regression(PCR) and Partial Least Squares Regression(PLSR) are the two most popular regression techniques in chemometrics. In the field of chemometrics usually the number of regressor variables greatly exceeds the number of observation. So we have to reduce the number of regressors to avoid the identifiability problem. In this paper we compare PCR and PLSR techniques combined with various robust regression methods including regression depth estimation. We compare the efficiency, goodness-of-fit and robustness of each estimators under several contamination schemes.

• Communications for Statistical Applications and Methods
• /
• 제18권6호
• /
• pp.761-770
• /
• 2011

회귀모형에 연관성이 높은 설명변수들이 포함되면 다중공선성의 문제가 야기되며, 동시에 자료에 회귀 이상점들이 포함되면 최소자승추정량에 바탕을 둔 제반 통계적 추론은 심각한 결함을 갖게 된다. 이러한 현상들은 데이터마이닝 분야에서 많이 볼 수 있는데, 본 논문에서는 두 가지 문제를 동시에 해결하기 위한 방안으로서 로버스트주성분회귀를 제안하였다. 특히 최적의 주성분을 선정하기 위한 새로운 기준을 개발하였는데, 설명변수들의 표본공분산 대신에 MVE-추정량을 기반으로 하였으며, 고유치가 아니라 상태지수의 크기에 바탕을 둔 선정기준을 제안하였다. 그리고 주성분모형에서의 추정을 위하여 회귀이상점에 대해 로버스트한 LTS-추정을 도입하였다. 제안된 선정기준이 기존의 기준들보다 다중공선성과 이상점이 유발하는 문제들을 잘 해결할 수 있음을 모의실험을 통하여 확인하였다.

• Journal of the Korean Data and Information Science Society
• /
• 제28권5호
• /
• pp.1191-1204
• /
• 2017

Weighted least squares (WLS) estimation is often easily used for the data with heteroscedastic errors because it is intuitive and computationally inexpensive. However, WLS estimator is less robust to a few outliers and sometimes it may be inefficient. In order to overcome robustness problems, Box-Cox transformation, Huber's M estimation, bisquare estimation, and Yohai's MM estimation have been proposed. Also, more efficient estimations than WLS have been suggested such as Bayesian methods (Cepeda and Achcar, 2009) and semiparametric methods (Kim and Ma, 2012) in heteroscedastic error models. Recently, Çelik (2015) proposed the weight methods applicable to the heteroscedasticity patterns including butterfly-distributed residuals and megaphone-shaped residuals. In this paper, we review heteroscedastic regression estimators related to robust or efficient estimation and describe their properties. Also, we analyze cost data of U.S. Electricity Producers in 1955 using the methods discussed in the paper.

• Journal of the Korean Data and Information Science Society
• /
• 제11권2호
• /
• pp.201-215
• /
• 2000

The classical method for regression analysis is the least squares method. However, if the data contain significant outliers, the least squares estimator can be broken down by outliers. To remedy this problem, the robust methods are important complement to the least squares method. Robust methods down weighs or completely ignore the outliers. This is not always best because the outliers can contain some very important information about the population. If they can be detected, the outliers can be further inspected and appropriate action can be taken based on the results. In this paper, I propose a sequential outlier test to identify outliers. It is based on the nonrobust estimate and the robust estimate of scatter of a robust regression residuals and is applied in forward procedure, removing the most extreme data at each step, until the test fails to detect outliers. Unlike other forward procedures, the present one is unaffected by swamping or masking effects because the statistics is based on the robust regression residuals. I show the asymptotic distribution of the test statistics and apply the test to several real data and simulated data for the test to be shown to perform fairly well.

• 응용통계연구
• /
• 제21권6호
• /
• pp.981-996
• /
• 2008

서로 다른 출처로부터 얻어진 데이터 파일들을 하나의 데이터 파일로 만드는 통계적 자료결합방법은 공통변수와 서로 다른 고유변수를 포함하여 변수들 간에 존재하는 관련성에 대해 살펴볼 수 있다. Robin (1986)이 제안한 일반회귀모형의 예측값을 이용한 통계적 결합방법은 자료에 대한 다변량 정규성을 가정하기 때문에 이 가정을 위반하는 자료를 이용하는 것은 많은 문제를 수반한다. 본 연구는 제공파일의 고유변수에 모분포를 반영하지 못하는 특이점이 존재하는 경우, 일반회귀모형을 이용한 통계적 결합방법의 대안으로 로러스트 회귀추정방법을 이용한 자료결합방법을 제안하였다. 나아가 로버스트 회귀모형을 이용한 결합방법과 일반회귀모형을 이용한 결합방법에서의 상관관계 및 결정계수 보존에 관한 성능을 비교하기 위하여 모의실험을 수행하였다.

• 한국통계학회:학술대회논문집
• /
• 한국통계학회 2004년도 학술발표논문집
• /
• pp.205-210
• /
• 2004

A weighted self-tuning robust regression estimator (WSTE) has the high breakdown point for estimating regression parameters such as other well known high breakdown estimators. In this paper, we propose to obtain standard quantities like confidence intervals, and it is found to be superior to the other high breakdown regression estimators when a sample is contaminated

• Communications for Statistical Applications and Methods
• /
• 제19권1호
• /
• pp.193-211
• /
• 2012

This paper deals with robust nonparametric regression analysis when the regressors are functional random fields. More precisely, we consider $Z_i=(X_i,Y_i)$, $i{\in}\mathbb{N}^N$ be a $\mathcal{F}{\times}\mathbb{R}$-valued measurable strictly stationary spatial process, where $\mathcal{F}$ is a semi-metric space and we study the spatial interaction of $X_i$ and $Y_i$ via the robust estimation for the regression function. We propose a family of robust nonparametric estimators for regression function based on the kernel method. The main result of this work is the establishment of the asymptotic normality of these estimators, under some general mixing and small ball probability conditions.