• The Korean Journal of Applied Statistics
• /
• v.31 no.4
• /
• pp.517-528
• /
• 2018

We provide R scripts to detect outliers in multivariate data and visualization. Detecting outliers is provided using three approaches 1) Robust Mahalanobis distance, 2) High Dimensional data, 3) density-based approach methods. We use the following techniques to visualize detected potential outliers 1) multidimensional scaling (MDS) and minimal spanning tree (MST) with k-means clustering, 2) MDS with fviz cluster, 3) principal component analysis (PCA) with fviz cluster. For real data sets, we use MLB pitching data including Ryu, Hyun-jin in 2013 and 2014. The developed R scripts can be downloaded at "http://www.knou.ac.kr/~sskim/ddpoutlier.html" (R scripts and also R package can be downloaded here).

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.5
• /
• pp.251-258
• /
• 2004

This paper proposes a method for estimating the illuminant chromaticity using the distributions of the camera responses obtained by a CCD camera in a real-world scene. Illuminant estimation using a highlight method is based on the geometric relation between a body and its surface reflection. In general, the pixels in a highlight region are affected by an illuminant geometric difference, camera quantization errors, and the non-uniformity of the CCD sensor. As such, this leads to inaccurate results if an illuminant is estimated using the pixels of a CCD camera without any preprocessing. Accordingly, to solve this problem the proposed method analyzes the distribution of the CCD camera responses and selects pixels using the Mahalanobis distance in highlight regions. The use of the Mahalanobis distance based on the camera responses enables the adaptive selection of valid pixels among the pixels distributed in the highlight regions. Lines are then determined based on the selected pixels with r-g chromaticity coordinates using a principal component analysis(PCA). Thereafter, the illuminant chromaticity is estimated based on the intersection points of the lines. Experimental results using the proposed method demonstrated a reduced estimation error compared with the conventional method.

• Korean Journal of Plant Resources
• /
• v.27 no.1
• /
• pp.89-94
• /
• 2014

Varietal distances were measured by Mahalanobis's $D^2$ statistics in 190 possible comparisons among twenty varieties of Prunus mume with twelve characters such as seed weight, length, width, and diameter, fruit weight, and number of sepals, petals, pistils, and stigmas, and leaf length and width. A complete linkage cluster analysis based on the Mahalanobis's distance ($D^2$) was attempt. Twenty varieties of Prunus mume were largely classified into five subgroups. Group I, II, III, IV and V included two, four, five, five and four varieties, respectively. Most of the varietal groups were not associated with their geographical origins. Number of stigmas, and leaf length and width among the twelve characters were the largest contributors to the $D^2$ in both intra-and inter groups.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.10 no.1
• /
• pp.84-90
• /
• 2011

In this paper, a VoIP-based voice secure telecommunication technology using the text-independent speaker authentication in the telematics environments is proposed. For the secure telecommunication, the sender's voice packets are encrypted by the public-key generated from the speaker's voice information and submitted to the receiver. It is constructed to resist against the man-in-the middle attack. At the receiver side, voice features extracted from the received voice packets are compared with the reference voice-key received from the sender side for the speaker authentication. To improve the accuracy of text-independent speaker authentication, Gaussian Mixture Model(GMM)-supervectors are applied to Support Vector Machine (SVM) kernel using Bayesian information criterion (BIC) and Mahalanobis distance (MD).

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.18 no.4
• /
• pp.393-399
• /
• 2008

The reduction of the Vehicle interior noise has been the main interest of NVH engineers. The driver's perception on the vehicle noise is affected largely by psychoacoustic characteristic of the noise as well as the SPL. The previous methods to evaluation of the SQ about vehicle interior noise are linear regression analysis of subjective SQ metrics by statistics and the estimation of the subjective SQ values by neural network. But these are so depended on jury test very much that they result in many difficulties. So, to reduce jury test weight, we suggested a new method using Mahalanobis distance for SQ evaluation. And, optimal characteristic values influenced on the result of the SQ evaluation were derived by signal to noise ratio(SN ratio) of the Taguchi method. Finally, the new method to evaluate SQ is constructed using Mahalanobis-Taguchi system(MTS). Furthermore, the MTS method for SQ evaluation was compared by the result of SQ grade table at the previous study and their virtues and faults introduced.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.1 s.307
• /
• pp.9-16
• /
• 2006

RDA(Relational Discriminant Analysis) is a way of finding classifiers based on the dissimilarity measures among the prototypes extracted from feature vectors instead of the feature vectors themselves. Therefore, the accuracy of the RDA classifier is dependent on the methods of selecting prototypes and measuring proximities. In this paper we propose to utilize PRS(Prototype Reduction Schemes) and Mahalanobis distances to devise a method of increasing classification accuracies. Our experimental results demonstrate that the proposed mechanism increases the classification accuracy compared with the conventional approaches for samples involving real-life data sets as well as artificial data sets.

• Korean Journal of Remote Sensing
• /
• v.31 no.4
• /
• pp.293-302
• /
• 2015

Water body extraction is significant for flood disaster monitoring using satellite imagery. Conventional methods have focused on finding an index, which highlights water body and suppresses non-water body such as vegetation or soil area. The Normalized Difference Water Index (NDWI) is typically used to extract water body from satellite images. The drawback of NDWI, however, is that some man-made objects in built-up areas have NDWI values similar to water body. The objective of this paper is to propose a new method that could extract correctly water body with built-up areas in before and after images of flood. We first create a two-element feature vector consisting of NDWI and a Near InfRared band (NIR) and then select a training site on water body area. After computing the mean vector and the covariance matrix of the training site, we classify each pixel into water body based on Mahalanobis distance. We also register before and after images of flood using outlier removal and triangulation-based local transformation. We finally create a change map by combining the before-flooding water body and after-flooding water body. The experimental results show that the overall accuracy and Kappa coefficient of the proposed method were 97.25% and 94.14%, respectively, while those of the NDWI method were 89.5% and 69.6%, respectively.

• Geomechanics and Engineering
• /
• v.35 no.5
• /
• pp.499-509
• /
• 2023

The objective of this study is to perform outlier analysis to obtain the distribution of groundwater levels through the best model. The groundwater levels are measured in 10, 25 and 30 piezometers in Seoul, Daejeon and Suncheon in South Korea. Fifty-eight empirical distribution functions were applied to determine a suitable fit for the measured groundwater levels. The best fitted models based on the measured values are determined as the Generalized Pareto distribution, the Johnson SB distribution and the Normal distribution for Seoul, Daejeon and Suncheon, respectively; the reliability is estimated through the Anderson-Darling method. In this study, to choose the appropriate confidence interval, the relationship between the amount of outlier data and the confidence level is demonstrated, and then the 95% is selected at a reasonable confidence level. The best model shows a smaller error ratio than the GEV while the Mahalanobis distance and outlier labelling methods results are compared and validated. The outlier labelling and Mahalanobis distance based on median shown higher validated error ratios compared to their mean equivalent suggesting, the methods sensitivity to data structure.

• Korean Journal of Remote Sensing
• /
• v.32 no.5
• /
• pp.471-482
• /
• 2016

This paper presents an extraction method for water body which uses block-based image partitioning and extension of water body boundaries to improve the performance of supervised classification for water body extraction. The Mahalanobis distance image is created by computing the spectral information of Normalized Difference Water Index (NDWI) and Near Infrared (NIR) band images over a training site within the water body in order to extract an initial water body area. To reduce the effect of noise contained in the Mahalanobis distance image, we apply mean curvature diffusion to the image, which controls diffusion coefficients based on connectivity strength between adjacent pixels and then extract the initial water body area. After partitioning the extracted water body image into the non-overlapping blocks of same size, we update the water body area using the information of water body belonging to water body boundaries. The update is performed repeatedly under the condition that the statistical distance between water body area belonging to water body boundaries and the training site is not greater than a threshold value. The accuracy assessment of the proposed algorithm was tested using KOMPSAT-2 images for the various block sizes between $11{\times}11$ and $19{\times}19$. The overall accuracy and Kappa coefficient of the algorithm varied from 99.47% to 99.53% and from 95.07% to 95.80%, respectively.

• Management Science and Financial Engineering
• /
• v.14 no.2
• /
• pp.1-12
• /
• 2008

This study explores several donor location strategies and discusses experiment results, which contributes to the saving of time and effort required in designing data fusion processes. In particular, three concepts are introduced. The Mahalanobis distance is applied to locate the nearest neighbors more effectively; which incorporates the covariance structure of attributes. The ideal point helps reduce the dimensionality problem that arises in conjoint-type experiments. The correspondence analysis is used to derive the coordinates from non-metric attributes. The Monte Carlo simulation results show that the proposed donor location strategies provide better fusion performance, compared to the currently-in-use methods.