• Journal of the Korean Data and Information Science Society
• /
• v.11 no.2
• /
• 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.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.7 no.3
• /
• pp.132-136
• /
• 2014

We introduce a new architecture of TSK-based fuzzy inference system. The proposed model used fuzzy c-means clustering method(FCM) for efficient disposal of data. The premise part of fuzzy rules don't assume any membership function such as triangular, gaussian, ellipsoidal because we construct the premise part of fuzzy rules using FCM. As a result, we can reduce to architecture of model. In this paper, we are able to use four types of polynomials as consequence part of fuzzy rules such as simplified, linear, quadratic, modified quadratic. Weighed Least Square Estimator are used to estimates the coefficients of polynomial. The proposed model is evaluated with the use of Boston housing data called Machine Learning dataset.

• Journal of the Korean Data and Information Science Society
• /
• v.6 no.1
• /
• pp.1-11
• /
• 1995

The parameters in the multiplicative model $Y_{1}={\alpha}_{0}{\prod}^{p}_{k=1}X_{kj}^{{\beta}_K}v_{j}$ are usually estimated by the least squares method after logarithmic transformation, and the least square Estimator of ${\alpha}_{0}$ is known to be biased, i.e., $E(e xp(\hat{\beta}_{0})){\neq}{\alpha}_{0})$. In the present study the unbaised estimators of ${\alpha}_{0}$ are examined(1) by modifying the least squares estimator and (2) by applying the Finney's results. The variances are also compared. In addition it has been observed that multiplicative model can be used to express the relationship beetween rice yield and yield components.

• Communications for Statistical Applications and Methods
• /
• v.4 no.3
• /
• pp.903-909
• /
• 1997

The ordinary least squares estimator and the corresponding pivotal statistics have been widely used for the unit test. Recently several test criteria based on maximum likelihood estimators and weighted symmetric estimator have been proposed for testing the unit root hypothesis in the autoregressive processes. Pantula at el. (1994) showed that the weighted symmetric estimator has good power properties. In this article we use an adjusted estimator for mean in the model when we use weighted symmetric estimator. A simulation study shows that for the small samples, this new test criterion has better power properties than the weighted symmetric estimator.

• Journal of the Korea Society of Computer and Information
• /
• v.10 no.2 s.34
• /
• pp.21-30
• /
• 2005

In general, the robust estimation method is well known for a good statistical estimator that is insensitive to small departures from the idealized assumptions for which the estimation is optimized. While there are many existing robust estimation techniques that have been proposed in the literature, two main techniques used in computer vision are M-estimators and least-median of squares (LMS). Among these. we utilized the M-estimators since they are known to provide an optimal estimation of affine motion parameters. The M-estimators have higher statistical efficiency but tolerate much lower percentages of outliers unless properly initialized. To resolve these problems, we proposed an adaptive M-estimators algorithm that effectively separates outliers from non-outliers and estimate affine model parameters, using a continuous sigmoid weight function. The experimental results show the superiority of our method.

• Communications for Statistical Applications and Methods
• /
• v.14 no.1
• /
• pp.229-239
• /
• 2007

We present a diagnostic method for the quasi-likelihood estimators in generalized linear models. Since these estimators can be usually obtained by iteratively reweighted least squares which are well known to be very sensitive to unusual data, a diagnostic step is indispensable to analysis of data. We extend the local influence approach based on the maximum likelihood function to that on the quasi-likelihood function. Under several perturbation schemes local influence diagnostics are derived. An illustrative example is given and we compare the results provided by local influence and deletion.

• Communications for Statistical Applications and Methods
• /
• v.3 no.3
• /
• pp.75-81
• /
• 1996

Polynomial measurement error model(MEM) with one predictor is considered. It is briefly mentioned that Chan and Mak's generalized least squares estimator(GLSE) can be derived more easily if Hermite polynomial concept is applied. It is proved that GLSE derived using new procedure is equivalent to the estimator obtained from corrected score function. Finally, much simpler proof of the asymptotic behavior of GLSE than that of Chan and Mak is provided. Much simpler formula of asymptotic covariance matrix of GLSE is a part of that proof.

• Journal of the Korean Statistical Society
• /
• v.25 no.2
• /
• pp.235-241
• /
• 1996

The ordinary least squares estimator of the disturbance variance in the linear regression model with stationary errors is shown to be asymptotically unbiased when the error process has a spectral density bounded from the above and away from zero. Such error processes cover a broad class of stationary processes, including ARMA processes.

• Communications for Statistical Applications and Methods
• /
• v.17 no.3
• /
• pp.349-356
• /
• 2010

The ordinary least squares based estimator of the disturbance variance in a regression model for spatial panel data is shown to be asymptotically unbiased and weakly consistent in the context of SAR(1), SMA(1) and SARMA(1,1)-disturbances when there is measurement error in the regressor matrix.

• The Korean Journal of Applied Statistics
• /
• v.26 no.3
• /
• pp.511-520
• /
• 2013

Detection outliers and robust estimators are crucial in regression models with outliers. In such studies the focus is on detecting outliers and estimating the coefficients using leave-one-out. Our study introduces Rice estimator which is an error variance estimator without estimating the coefficients. In particular, we study a comparison of the statistical properties for Rice estimator with and without outliers in simple regression models.