• Communications for Statistical Applications and Methods
• /
• v.3 no.2
• /
• pp.21-28
• /
• 1996

Collinearities contained in MEM cause the same problems as they do in traditional regression model, so the detection of collinearities is a crucial topic in MEM. One diagnostic was introduced by Carrillo-Gamboa and Gunst, but their method did not work in some cases. Two alternative collinearity diagnostics that provide reasonable measure of collinearities are proposed. Simulation study is performed to compare the small-sample properties of the proposed collinearity diagnostics.

• Proceedings of the KSRS Conference
• /
• v.2
• /
• pp.578-581
• /
• 2006

Usually to achieve precise geolocation of satellite images, we need to get GCPs (Ground control points) from individual scenes. This requirement greatly increases the cost and processing time for satellite mapping. In this article, we focus on finding appropriate sensor models for entire image strips composing of several adjacent scenes. We tested the feasibility of modelling whole satellite image strips by establishing sensor models of one scene with GCPs and by applying the models to neighboring scenes without GCPs. For this, we developed two types of sensor models: collinearity-based type and orbit-based type and tested them using different sets of unknowns. Results indicated that although the performance of two types was very similar, for modelling individual scenes, it was not for modelling the whole strips. Moreover, the performance of sensor models was remarkably sensitive to different sets of unknowns. It was found that the orbit-based model using attitude biases as unknowns can be used to model SPOT image strips of 420 Km in length.

• Bulletin of the Korean Chemical Society
• /
• v.22 no.1
• /
• pp.37-42
• /
• 2001

Ridge regression is compared with multiple linear regression (MLR) for determination of Research Octane Number (RON) when the baseline and signal-to-noise ratio are varied. MLR analysis of near-infrared (NIR) spectroscopic data usually encounters a collinearity problem, which adversely affects long-term prediction performance. The collinearity problem can be eliminated or greatly improved by using ridge regression, which is a biased estimation method. To evaluate the robustness of each calibration, the calibration models developed by both calibration methods were used to predict RONs of gasoline spectra in which the baseline and signal-to-noise ratio were varied. The prediction results of a ridge calibration model showed more stable prediction performance as compared to that of MLR, especially when the spectral baselines were varied. . In conclusion, ridge regression is shown to be a viable method for calibration of RON with the NIR data when only a few wavelengths are available such as hand-carry device using a few diodes.

• Journal of the Korean Statistical Society
• /
• v.28 no.3
• /
• pp.325-336
• /
• 1999

Collinearity among independent variables can have severe effects on the precision of response estimation for some region of interest in the experiments with mixture. A method of finding optimal linear restriction on regression parameter in linear model for mixture experiments in the sense of minimizing integrated mean squared error is studied. We use the formulation of optimal restrictions on regression parameters for estimating responses proposed by Park(1981) by transforming mixture components to mathematically independent variables.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.28 no.3
• /
• pp.61-66
• /
• 2003

In many multiple regression analyses, the “multi-collinearity” problem arises since some independent variables are highly correlated with each other. Practically, the Ridge regression method is often adopted to deal with the problems resulting from multi-collinearity. We propose a better alternative method using iteration to obtain an exact least squares estimator. We prove the solvability of the proposed algorithm mathematically and then compare our method with the traditional one.

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.1403-1405
• /
• 2003

This paper focuses on the image correction under few GCPs, utilizes the collinearity equation, and builds up this mathematics model of space backside resection based on condition adjustment. Then calculates the adjusted elements of exterior orientation by iteration algorithm, and evaluates the precision. And demonstrates the high-precision, affection and wide-supplying-perspective of this model.

• Korean Journal of Geomatics
• /
• v.4 no.1
• /
• pp.31-37
• /
• 2004

The aim of this research is comparing the existing approximation models (e.g. Affine Transformation and Direct Linear Transformation) with Rational Function Model as a substitute of rigorous sensor model of linear array scanner, especially push-broom sensor. To do so, this research investigates the mathematical model of each approximation method. This is followed by the assessments of accuracy of transformation from object space to image space by using simulated data generated by collinearity equations which incorporate or depict the physical aspects of linear array sensor.

• Journal of Korean Society for Atmospheric Environment
• /
• v.21 no.3
• /
• pp.315-327
• /
• 2005

The total of 35 target VOCs (volatile organic compounds), which were included in the TO-14, was selected to develop a VOCs' source profile matrix of the Yeosu Petrochemical Complex and to test its collinearity by singular value decomposition(SVD) technique. The VOCs collected in canisters were sampled from 12 different sources such as 8 direct emission sources (refinery, painting, wastewater treatment plant, incinerator, petrochemical processing, oil storage, fertilizer plant, and iron mill) and 4 general area sources (gasoline vapor emission, graphic art activity, vehicle emission, and asphalt paving activity) in this study area, and then those samples were analyzed by GC/MS. Initially the resulting raw data for each profile were scaled and normalized through several data treatment steps, and then specific VOCs showing major weight fractions were intensively reviewed and compared by introducing many other related studies. Next, all of the source profiles were tested in terms of degree of collinearity by SVD technique. The study finally could provide a proper VOCs' source profile in the study area, which can give opportunities to apply various receptor models properly including chemical mass balance (CMB).

• Proceedings of the KSRS Conference
• /
• 2005.10a
• /
• pp.91-94
• /
• 2005

A numerical formula that presents relationship between a point of a satellite image and its ground position is called a sensor model. For precise geolocation of satellite images, we need an error-free sensor model. However, the sensor model based on GOES ephemeris data has some error, in particular after Image Motion Compensation (IMC) mechanism has been turned off. To solve this problem, we investigate three sensor models: Collinearity model, Direct Linear Transform (DLT) model and Orbit-based model. We apply matching between GOES images and global coastline database and use successful results as control points. With control points we improve the initial image geolocation accuracy using the three models. We compare results from three sensor models that are applied to GOES-9 images. As a result, a suitable sensor model for precise geolocation of GOES-9 images is proposed.

• Communications for Statistical Applications and Methods
• /
• v.21 no.5
• /
• pp.395-409
• /
• 2014

Data limited, partial, or incomplete are known as an ill-posed problem. If the data with ill-posed problems are analyzed by traditional statistical methods, the results obviously are not reliable and lead to erroneous interpretations. To overcome these problems, we propose a dual generalized maximum entropy (dual GME) estimator for panel data regression models based on an unconstrained dual Lagrange multiplier method. Monte Carlo simulations for panel data regression models with exogeneity, endogeneity, or/and collinearity show that the dual GME estimator outperforms several other estimators such as using least squares and instruments even in small samples. We believe that our dual GME procedure developed for the panel data regression framework will be useful to analyze ill-posed and endogenous data sets.