• Communications for Statistical Applications and Methods
• /
• v.21 no.5
• /
• pp.445-459
• /
• 2014

This paper investigates the dependence structure of Korean financial markets (stock, foreign exchange (FX) rates and bond) using copula-GARCH and dynamic conditional correlation (DCC) models. We examine GJR-GARCH with skewed elliptical distributions and four copulas (Gaussian, Student's t, Clayton and Gumbel) to model dependence among returns, and then employ DCC model to describe system-wide correlation dynamics. We analyze the daily returns of KOSPI, FX (WON/USD) and KRX bond index (Gross Price Index) from $2^{nd}$ May 2006 to $30^{th}$ June 2014 with 2,063 observations. Empirical result shows that there is significant asymmetry and fat-tail of individual return, and strong tail-dependence among returns, especially between KOSPI and FX returns, during the 2008 Global Financial Crisis period. Focused only on recent 30 months, we find that the correlation between stock and bond markets shows dramatic increase, and system-wide correlation wanders around zero, which possibly indicates market tranquility from a systemic perspective.

• Environmental and Resource Economics Review
• /
• v.9 no.1
• /
• pp.71-91
• /
• 1999

상수원이 오염되고 사람들이 수돗물 수질을 불신함에 따라 이에 대한 회피 행동(averting behavior)으로서 생수소비가 급증하고 있다. 본 논문은 모수적(parametric) 모형과 반모수적(semi-parametric) 모형을 운용하여 생수소비 행태를 설명하는 이변량모형(bivariate model)을 분석하고자 한다. 주된 분석목적은 모수적 모형과 관련된 암묵적인 제약이 반모수적 모형을 통해 완화될 수 있음을 보이는 것이다. 모수적 모형에서 가정되는 이변량 정규성(normality)의 가정에 대한 검정결과, 이 가정은 통계적으로 유의하게 기각되었으며 모수적 모형 대반모수적 모형에 대한 정형검정 결과도 모수적 모형을 유의하게 기각하였다. 또한 반모수적 모형이 모수적 모형보다 생수소비행태에 대한 설명력이 더 높았다. 따라서, 반모수적 모형이 모수적 모형에서 요구되는 제약적인 가정을 하지 않으면서도 자료와 보다 잘 부합하는 결과를 가져올 수 있는 것으로 분석되었다.

• Journal of The Korean Astronomical Society
• /
• v.46 no.6
• /
• pp.261-268
• /
• 2013

Magnetohydrostatic equilibria, in which the Lorentz force, the plasma pressure force and the gravitational force balance out to zero, are widely adopted as the zeroth order states of many astrophysical plasma structures. A magnetic flux-current surface is a surface, in which both magnetic field lines and current lines lie. We for the first time derive the necessary and sufficient condition for existence of magnetic flux-current surfaces in magnetohydrostatic equilibria. It is also shown that the existence of flux-current surfaces is a necessary (but not sufficient) condition for the ratio of gravity-aligned components of current density and magnetic field to be constant along each field line. However, its necessary and sufficient condition is found to be very restrictive. This finding gives a significant constraint in modeling solar coronal magnetic fields as force-free fields using photospheric magnetic field observations.

• International Journal of Precision Engineering and Manufacturing
• /
• v.9 no.2
• /
• pp.3-7
• /
• 2008

Vibratory conveyors are widely used in industry to transport granular materials and products. A theoretical point mass model for vibratory conveying was studied. The results agreed well with experimental observations. The model theory included the resting, sliding and flight states of the material. Each state was considered separately when determining the equations of motion. For the coefficients of restitution, values of zero for the normal component and 0.8 for the tangential component were found to be appropriate for modeling the collisions of the granular particles with the conveying surface. The vibration angle had a large influence on the mode and rate of transport. There was an optimum vibration angle for a given set of conditions. The optimum vibration angle decreased and was better defined as the coefficient of friction increased. The results suggest the existence of an optimum dimensionless track acceleration (throw number), which does not support general industrial practice in which the track acceleration is limited when the feed cycle becomes erratic and unstable.

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.257-259
• /
• 2003

Landuse maps are prepared from satellite imagery and field observations were conducted at various locations in the study area. Compared to the field data and NDVI and RVI thematic maps, NDVI is better than RVI, because it compensates for changing illumination conditions, surface slope, aspect and other factors. Clouds, water and snow have negative values for RVI and NDVI. Rock and bare soils have similar reflectance in both NIR and visible band, so RVI and NDVI are near zero. In forest areas with good vegetation cover, NDVI is high and landslide occurrence is less. But if annual and biennial vegetations are present and if cultivation practices are changed frequently, NDVI is medium and landslide occurrence is moderate. In areas where deforestation and settlement is in progress, NDVI is less and landslide occurrence is more. The NDVI FCC thematic map may be used as an important layer in GIS application for landslide studies. Analyzing other layers such as slope, rainfall, soil, geology, drainage, lineament, etc with NDVI FCC layer will give a better idea about the identity of landslide prone areas.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.12 no.3
• /
• pp.193-197
• /
• 2012

This paper focuses on identifying which appliance is currently operating by analyzing electrical load signature for home energy monitoring system. The identification framework is comprised of three steps. Firstly, specific appliance features, or signatures, were chosen, which are DC (Duty Cycle), SO (Slope of On-state), VO (Variance of On-state), and ZC (Zero Crossing) by reviewing observations of appliances from 13 houses for 3 days. Five appliances of electrical rice cooker, kimchi-refrigerator, PC, refrigerator, and TV were chosen for the identification with high penetration rate and total operation-time in Korea. Secondly, K-NN and Naive Bayesian classifiers, which are commonly used in many applications, are employed to estimate from which appliance the signatures are obtained. Lastly, one of candidates is selected as final identification result by majority voting. The proposed identification frame showed identification success rate of 94.23%.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.1
• /
• pp.58.3-58.3
• /
• 2019

Optical intensity interferometry, developed in the 1950s, is a simple and inexpensive method for achieving angular resolutions on microarcsecond scales. Its low sensitivity has limited intensity interferometric observations to bright stars so far. Substantial improvements are possible by using avalanche photodiodes (APDs) as light detectors. We present here the results of laboratory measurements with a prototype astronomical intensity interferometer using APDs in continuous ("linear") detection mode - arguably, the first of its kind. We used two interferometer configurations, one with zero baseline and one with variable baseline. Using a superluminous diode as light source, we unambiguously detected Hanbury Brown-Twiss photon-photon correlations at very high significance. From measuring the correlation as function of baseline, we measured the angular diameter of the light source, in analogy to the measurement of the angular diameter of a star. Our results demonstrate the possibility to construct large astronomical intensity interferometers that can address a multitude of astrophysical science cases.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.2
• /
• pp.53.3-53.3
• /
• 2019

Recent work by Ravanbakhsh et al. (2017), Mathuriya et al. (2018) showed that convolutional neural networks (CNN) can be trained to predict cosmological parameters from the visual shape of the large scale structure, i.e. the filaments, clusters and voids of the cosmic density field. These preliminary works used the dark matter density field at redshift zero. We build upon these works by considering realistic mock galaxy catalogues that mimic true observations. We construct light-cones that span the redshift range appropriate for current and near future cosmological surveys such as LSST, EUCLID, WFIRST etc. In summary, we propose a novel multi-image input CNN to track the evolution in the morphology of large scale structures over cosmic time to constrain cosmology and the expansion history of the Universe.

• Journal of Korean Society for Geospatial Information Science
• /
• v.23 no.4
• /
• pp.11-16
• /
• 2015

Spatial and temporal information on near-surface air temperature is important for understanding global warming and climate change. In this study, the estimation algorithm of near-surface air temperature in Korea was developed by using spatial homogeneous surface information obtained from satellite remote sensing observations. Based on LST(Land Surface Temperature), NDWI(Normalized Difference Water Index) and NDVI(Normalized Difference Vegetation Index) as independent variables, the multiple regression model was proposed for the estimation of near-surface air temperature. The different regression constants and coefficients for warm and cold seasons were calculated for considering regional climate change in Korea. The near-surface air temperature values estimated from the multiple regression algorithm showed reasonable performance for both warm and cold seasons with respect to observed values (approximately $3^{\circ}C$ root mean-square error and nearly zero mean bias). Thus;the proposed algorithm using remotely sensed surface observations and the approach based on the classified warm and cold seasons may be useful for assessment of regional climate temperature in Korea.

• Journal of Distribution Science
• /
• v.14 no.10
• /
• pp.109-116
• /
• 2016

Purpose - This study investigates the purchase pattern relating to soybean products in Korea. Specifically, the effect of branding based on a regional name was analyzed in terms of consumer purchase frequencies. The primary purpose of this study is to understand why family characteristics affect product selection for a regional brand in the soybean food category. Research design, data, and methodology - We used data collected by the Rural Development Administration (RDA) of Korea. The RDA has monitored agricultural food consumers for years in order to obtain purchase records. Panel participants live in regions near the capital city of Seoul, Korea. Examining data from January 2010 to May 2016, 667 families were selected for analysis. The final data set was 1,335,402. Each purchase item by each individual family was aggregated to a countable weekly observation. To analyze the data set quantitatively, zero-inflation regression was adopted, which was appropriate to avoid biases from overly dispersed observations. Results - We hypothesized the effects of regional branding from the viewpoint of the family characteristics. The first hypothesis was that the number of children would be positively associated with the purchase of a regional brand of soybean products. The result strongly supported this hypothesis. The second hypothesis was that the number of family members would be negatively associated with the purchase of the soybean products of a regional brand. Based on empirical analysis, we concluded that this hypothesis was partially supported. The third hypothesis was the presence of an interaction effect between the number of children and the family size, which was supported by the results. As a supplementary analysis, we also tested mean-variance differences in terms of categories and regional branding with corporate branding. Conclusion - The results of this study provide insights for regional branding strategies in agricultural food management. This study appears to be one of the seminal studies trying to analyze purchase patterns from longitudinal observations. In addition, this study adopted variables characterizing family lifestyle. This study confirmed that children and family size should be considered when soybean product brands are introduced.