• 한국물환경학회지
• /
• 제33권6호
• /
• pp.650-660
• /
• 2017

In this study, pollutant emission characteristics by water damage period analyzed 11 items (water temperature, pH, DO, EC, BOD, COD, TOC, SS, T-N, T-P and flow) with load duration curve, time series load curve and factor analysis for three years (2014-2016). Load duration curve is applied to judge the level of impaired waterbody and estimate impaired level by pollutants such as BOD and T-P in this study depending on variation of stream flow. Water quality standard exceeded the flow of mid-range and low-range by flow condition evaluation using load duration curve. This watershed was influenced by point source more than non-point source. Cumulative excess rate of BOD and T-P kept water quality standard for all seasons (spring, summer, autumn and winter) except BOD 59% in spring. Water quality changes were influenced by pollutants of basic environmental treatment facilities and agricultural areas during spring and summer. Results of factor analysis were classified commonly first factor (BOD, COD, and TOC) and second factor (flow, water temperature and SS). Therefore, effects of artificial pollutants and maintenance water must be controlled seasonally and reduced relative to water damage caused by point pollution sources with effluent standard strengthened in the target watershed.

• 한국항해항만학회지
• /
• 제46권4호
• /
• pp.385-391
• /
• 2022

본 연구에서는 수요 및 공급요인 외 해상운임에 영향을 주는 다양한 변수들을 발굴하고자 다변량 시계열분석을 수행해 보았다. 우선 종속변수에는 해상운임을 대용할 변수로 Shipping Intelligence에서 제공하고 있는 종합운임지수(ClarkSea Index), 벌크선운임(Clarksons Average Bulker Earnings), 탱커선운임(Clarksons Average Tanker Earnings) 등을 활용하였다. 선행연구를 통해 해상운임에 영향을 미칠 것으로 예상되는 세계 해상물동량(World Seaborne Trade), 세계 선복량(World Fleet), 유가(Brent Crude Oil Price), 세계 GDP성장률(GDP World), OECD 산업생산성장률(Industrial Production OECD), 금리(US$ LIBOR 6 Months), OECD 인플레이션(CPI OECD) 등을 독립변수로 설정하여 회귀분석을 수행해 보았다. 데이터는 시계열자료로 1992년부터 2020년까지의 연데이터로 구성하였다. 분석결과 종합운임지수에는 해상물동량과 유가가, 벌크선운임에는 해상물동량만이, 탱커선운임에는 해상물동량, 유가, 산업생산성장률, 인플레이션 등이 통계적으로 유의미한 영향을 미치는 것으로 나타났다.

• 한국전자통신학회논문지
• /
• 제19권2호
• /
• pp.383-388
• /
• 2024

본 논문은 전라남도에서 측정한 태양광 발전 데이터를 기반으로 발전량 예측값을 도출하기 위한 연구이다. 발전량 측정을 위해 인버터에서 직류, 교류, 환경데이터와 같은 다변량 변수를 측정하였고, 측정값의 안정성과 신뢰성 확보를 위한 전처리 작업을 수행하였다. 상관관계 분석은 부분자기상관함수(PACF: Partial Autocorrelation Function)을 활용하여 시계열 데이터에서 발전량과 상관성이 높은 데이터만을 예측을 위해 사용하였다. 태양광 발전량 예측을 위해 딥러닝 모델을 이용하여 발전량을 측정했고, 예측 정확도를 높이기 위해 각 다변량 변수의 상관관계 분석 결과를 이용하였다. 정제된 데이터를 활용한 학습은 기존 데이터를 그대로 사용했을 때 보다 안정되었고, 상관관계 분석 결과를 반영하여 다변량 변수 중 상관성이 높은 변수만을 활용하여 태양광 발전량 예측 알고리즘을 개선하였다.

• 농촌지도와개발
• /
• 제2권1호
• /
• pp.1-21
• /
• 1995

This study examines the socioeconomic returns to agricultural research and extension services in Korea, using multivariate time series technique and Akino-Hyami formula. Results find that the socioeconomic returns are quite competitive in case of agricultural research with 44.82% internal rate of return and very high in case of extension services with 207.82% internal rate of return. Agricultural production responds to the agricultural research shock about four years after the shock. The magnitudes of the impacts increase until a peak is reached nine years after the initial expenditures and the impacts declines to a zero level after about twenty years. This lag lengths are consistent to the usual literature on research, which finds lags of seven to thirty years. Agricultural production responds to an agricultural extension shock immediately and declines to a zero level after about four years. Thus, the lag lengths are much shorter than those by research shock.

• Communications for Statistical Applications and Methods
• /
• 제22권6호
• /
• pp.615-624
• /
• 2015

We consider the estimation of seasonal cointegration in the presence of conditional heteroskedasticity (CH) using a feasible generalized least squares method. We capture cointegrating relationships and time-varying volatility for long-run and short-run dynamics in the same model. This procedure can be easily implemented using common methods such as ordinary least squares and generalized least squares. The maximum likelihood (ML) estimation method is computationally difficult and may not be feasible for larger models. The simulation results indicate that the proposed method is superior to the ML method when CH exists. In order to illustrate the proposed method, an empirical example is presented to model a seasonally cointegrated times series under CH.

• 가정과삶의질연구
• /
• 제18권1호
• /
• pp.167-184
• /
• 2000

This study has investigated the degree of similarities and/or differences of consumption expenditure styles among the households with different occupation. Two types of analysis were performed. One was comparative analysis which used to identify the changes of consumption expenditure styles among different occupation classes using time-series data of 1977-1996 Korean Urban Household Expenditure Survey and Rural Household Economy Survey. The other was multivariate analysis to investigate the effects of occupation on consumption expenditure styles with 1551 sample household data from 1996 Korean Urban Household Expenditure Survey. The results showed that the differences in consumption styles among different occupation classes including farmers are diminished during last two decades while there still exist some degree of differences in consumption styles. After controlling other socio-economic factors it is found that occupation is a determinant of consumption patterns of urban w ge earners especially consumption for clothing and health items.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2003년도 봄 학술발표논문집 Vol.30 No.1 (B)
• /
• pp.178-180
• /
• 2003

EEG-based brain computer interface(BCI) provides a new communication channel between human brain and computer. EEG data is a multivariate time series so that hidden Markov model (HMM) might be a good choice for classification. However EEG is very noisy data and contains artifacts, so useful features mr expected to improve the performance of HMM. In this paper we addresses the usefulness of principal component features with Hidden Markov model (HHM). We show that some selected principal component features can suppress small noises and artifacts, hence improves classification performance. Experimental study for the classification of EEG data during imagination of a left, right up or down hand movement confirms the validity of our proposed method.

• 통상정보연구
• /
• 제10권3호
• /
• pp.459-474
• /
• 2008

Under violent competition to secure international raw materials, safe supply and demand of crude oil that only relies on import among main raw materials is an important task for Korean economic development. Therefore, this study aims to analyze the impact of price change of international crude oil on merchandise balance. It also presents political suggestions in preparation for national economic development and safety and develops an organized and long-term overseas resources development program. As the time-series data which had the 1st difference contribute to dismissal of the null hypothesis successfully, we carry out a multivariate cointegration test developed by Johansen (1988) and find that at least one cointegration vector exists. And, when Impulse Response Function is introduced, as the crude oil import price shows a negative impact from Step 2, then an extreme change, a positive impact since Step 13, is maintained and a safe result appears.

• Journal of the Korean Data and Information Science Society
• /
• 제19권4호
• /
• pp.1201-1210
• /
• 2008

The goal of this paper is to investigate the trend of stream quality and the quality of water in Nakdong river by the method of factor analysis. It used the fourteen different monthly time series data such as pH, BOD, COD, SS, TN and etc. of the thirty four of Nakdong River measurement points from Jan. 1998 to Dec. 2006. The result of factor analysis is that the factor 1 results from organic water pollution is occupied 29.288% such as BOD, COD, TN and EC, and the factor 2 explained from sewage and a seasonal variation is occupied 16.467% such as SS.

• Environmental Analysis Health and Toxicology
• /
• 제23권4호
• /
• pp.231-245
• /
• 2008

Various methods for detecting changes in response behaviors of indicator specimens are presented for monitoring effects of toxic treatments. The movement patterns of individuals are quantitatively characterized by statistical (i.e., ANOVA, multivariate analysis) and computational (i.e., fractal dimension, Fourier transform) methods. Extraction of information in complex behavioral data is further illustrated by techniques in ecological informatics. Multi-Layer Perceptron and Self-Organizing Map are applied for detection and patterning of response behaviors of indicator specimens. The recent techniques of Wavelet analysis and line detection by Recurrent Self-Organizing Map are additionally discussed as an efficient tool for checking time-series movement data. Behavioral monitoring could be established as new methodology in integrative ecological assessment, tilling the gap between large-scale (e.g., community structure) and small-scale (e.g., molecular response) measurements.