• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2020년도 학술발표회
• /
• pp.170-170
• /
• 2020

Due to the development of technologies, complex computation of huge data set is possible with a prevalent personal computer. Therefore, machine learning methods have been widely applied in the hydrologic field such as regression-based regional frequency analysis (RFA). The main purpose of this study is to compare two frameworks of RFA based on the artificial neural network (ANN) models: quantile regression technique (QRT-ANN) and parameter regression technique (PRT-ANN). As an output layer of the ANN model, the QRT-ANN predicts quantiles for various return periods whereas the PRT-ANN provides prediction of three parameters for the generalized extreme value distribution. Rainfall gauging sites where record length is more than 20 years were selected and their annual maximum rainfalls and various hydro-meteorological variables were used as an input layer of the ANN model. While employing the ANN model, 70% and 30% of gauging sites were used as training set and testing set, respectively. For each technique, ANN model structure such as number of hidden layers and nodes was determined by a leave-one-out validation with calculating root mean square error (RMSE). To assess the performances of two frameworks, RMSEs of quantile predicted by the QRT-ANN are compared to those of the PRT-ANN.

• 한국수자원학회논문집
• /
• 제45권8호
• /
• pp.815-826
• /
• 2012

기존 Ordinary Regression (OR) 방법을 이용한 경향성 분석은 경향성을 과소평가하는 문제점을 나타낸다. 이러한 점에서 본 연구에서는 자료의 정규분포 가정과 평균을 중심으로 경향성 평가가 이루어지는 기존 Ordinary Regression (OR) 방법을 개선한 Quantile Regression (QR) 방법을 제안하였다. 본 연구에서는 64개 강우 관측지점의 연 최대 극대강수량 자료에 대하여 QR 방법과 OR 방법에 대하여 통계적 성능을 평가하였다. QR 방법의경향성 분석결과 47개 지점에서 5% 오차수준 내에서 t-검정을 통과한 반면 OR 방법에서는 13개 지점 만이 통계적 유의성을 가지는 것으로 나타났다. 이는 OR 방법이 자료의 평균을 중심으로 경향성을 평가하는 기법인데 반해 QR은 자료의 다양한 분위에서 경향성을 평가함으로써 극대 및 극소 부분에서의 경향성을 보다 유연하게 감지하는 이유로 판단된다. QR 방법을 통한 경향성 평가는 평균 중심의 해석문제점을 개선할 수 있으며 자료가 정규분포를 따르지 않거나 왜곡된 분포형태를 갖는 자료의 수문학적 경향성 평가에 유용하게 사용될 수 있을 것으로 판단된다.

• 한국통계학회:학술대회논문집
• /
• 한국통계학회 2000년도 추계학술발표회 논문집
• /
• pp.187-191
• /
• 2000

In the analysis of repeated measurements, multivariate regression models that account for the correlations among the observations from the same subject are widely used. Like the usual univariate regression models, these multivariate regression models also need some model diagnostic procedures. In this paper, we propose a simple graphical method to detect outliers and to investigate the goodness of model fit in repeated measures data. The graphical method is based on the quantile-quantile(Q-Q) plots of the $X^2$ distribution and the standard normal distribution. We also propose diagnostic measures to detect influential observations. The proposed method is illustrated using two examples.

• 노동경제논집
• /
• 제28권1호
• /
• pp.135-156
• /
• 2005

본 연구에서는 자영업 부문의 소득분포 및 소득결정요인을 임금근로와 비교 분석하였다. 자영업 부문은 임금근로보다 소득편차가 크고, 부문내 이질성이 큰 집단이라는 점에서, OLS 추정과 더불어 분위회귀분석(quantile regression analysis)을 실시하였다. 주요 분석 결과를 보면, 첫째, 자영업주의 소득이 임금근로자보다 높으며, 소득분위가 높아질수록 자영업주와 임금근로자간의 소득격차가 확대된다. 둘째, 교육의 한계효과는 자영업주와 임금근로자 공히 소득분위가 높아질수록 증가하고 있어, 소득분위가 높은 집단일수록 교육에 대한 보상(가격)이 높다는 것을 알 수 있다. 단, 여성 자영업주의 경우는 예외로서, 소득분위가 높을수록 교육의 한계효과가 감소한다. 즉 소득분위가 높은 집단에 속하는 임금근로자와 남성 자영업주는 소득분위가 낮은 집단에 비해 노동시장에서 교육에 대한 보상이 더 큰 반면, 여성 자영업주는 소득분위가 높은 집단에서 교육에 대한 보상이 오히려 작다.

• Communications for Statistical Applications and Methods
• /
• 제15권5호
• /
• pp.753-764
• /
• 2008

The $L_1$ regularized estimator in quantile problems conduct parameter estimation and model selection simultaneously and have been shown to enjoy nice performance. However, $L_1$ regularized estimator has a drawback: when there are several highly correlated variables, it tends to pick only a few of them. To make up for it, the proposed method adopts doubly regularized framework with the mixture of $L_1$ and $L_2$ norms. As a result, the proposed method can select significant variables and encourage the highly correlated variables to be selected together. One of the most appealing features of the new algorithm is to construct the entire solution path of doubly regularized quantile estimator. From simulations and real data analysis, we investigate its performance.

• Asian Pacific Journal of Cancer Prevention
• /
• 제17권sup3호
• /
• pp.113-117
• /
• 2016

Quantile regression is an efficient method for predicting and estimating the relationship between explanatory variables and percentile points of the response distribution, particularly for extreme percentiles of the distribution. To study the relationship between urbanization and cancer morbidity, we here applied quantile regression. This cross-sectional study was conducted for 9 cancers in 345 cities in 2007 in Iran. Data were obtained from the Ministry of Health and Medical Education and the relationship between urbanization and cancer morbidity was investigated using quantile regression and least square regression. Fitting models were compared using AIC criteria. R (3.0.1) software and the Quantreg package were used for statistical analysis. With the quantile regression model all percentiles for breast, colorectal, prostate, lung and pancreas cancers demonstrated increasing incidence rate with urbanization. The maximum increase for breast cancer was in the 90th percentile (${\beta}$=0.13, p-value<0.001), for colorectal cancer was in the 75th percentile (${\beta}$=0.048, p-value<0.001), for prostate cancer the 95th percentile (${\beta}$=0.55, p-value<0.001), for lung cancer was in 95th percentile (${\beta}$=0.52, p-value=0.006), for pancreas cancer was in 10th percentile (${\beta}$=0.011, p-value<0.001). For gastric, esophageal and skin cancers, with increasing urbanization, the incidence rate was decreased. The maximum decrease for gastric cancer was in the 90th percentile(${\beta}$=0.003, p-value<0.001), for esophageal cancer the 95th (${\beta}$=0.04, p-value=0.4) and for skin cancer also the 95th (${\beta}$=0.145, p-value=0.071). The AIC showed that for upper percentiles, the fitting of quantile regression was better than least square regression. According to the results of this study, the significant impact of urbanization on cancer morbidity requirs more effort and planning by policymakers and administrators in order to reduce risk factors such as pollution in urban areas and ensure proper nutrition recommendations are made.

• 응용통계연구
• /
• 제33권5호
• /
• pp.569-578
• /
• 2020

분위수 회귀모형은 반응변수의 조건부 분위수 함수를 추정함으로써 반응변수와 예측변수의 관계에 대한 포괄적인 정보를 제공한다. 특히 커널 분위수 회귀모형은 비선형 관계식을 고려하기 위하여 양정치 커널함수(kernel function)에 의해 만들어지는 재생 커널 힐버트 공간(reproducing kernel Hilbert space)에서 비선형 조건부 분위수 함수를 추정한다. 그러나 KQR은 이차계획법으로 공식화되어 많은 계산비용을 필요로 하므로 컴퓨터 메모리 능력의 제한으로 대용량 자료의 분석은 불가능하다. 이러한 문제점을 해결하기 위하여 본 논문에서는 분할정복(divide and conquer) 알고리즘을 활용한 KQR 추정법(DC-KQR)을 제안한다. DC-KQR은 먼저 전체 훈련자료를 몇 개의 부분집합으로 무작위로 분할(divide)한 후, 각각의 부분집합에 대하여 KQR 분위수 함수를 추정하고 이들의 산술 평균을 이용하여 최종적인 추정량으로 통합(conquer)하는 기법이다. 본 논문에서는 모의실험과 실제자료 분석을 통해 제안한 DC-KQR의 효율적인 성능과 활용 가능성을 확인하였다.

• 응용통계연구
• /
• 제21권4호
• /
• pp.631-638
• /
• 2008

가속수명시험은 실제 사용조건보다 열악한 수준으로 시험하여 빠른 기간 내에 제품의 고장자료를 얻고, 실제 사용조건에서의 수명관련 품질 특성치를 추정하는 방법이다. 본 논문에서는 가속수명 자료를 이용하여 분위수 회귀추정 방법을 통해 정상 조건에서의 수명을 추정하는 방법을 제안한다. 대표적인 가속 스트레스인 온도와 전압을 갖는 실제 자료에 분위수 회귀 모형을 적용하여 수명을 추정하였다.

• Journal of Preventive Medicine and Public Health
• /
• 제56권6호
• /
• pp.515-522
• /
• 2023

Objectives: The pandemic caused by coronavirus disease 2019 (COVID-19) has exerted an unprecedented impact on the health of populations worldwide. However, the adverse health consequences of the pandemic in terms of infection and mortality rates have varied across countries. In this study, we investigate whether COVID-19 mortality rates across a group of developed nations are associated with characteristics of their healthcare systems, beyond the differential policy responses in those countries. Methods: To achieve the study objective, we distinguished healthcare systems based on the extent of healthcare decommodification. Using available daily data from 2020, 2021, and 2022, we applied quantile regression with non-additive fixed effects to estimate mortality rates across quantiles. Our analysis began prior to vaccine development (in 2020) and continued after the vaccines were introduced (throughout 2021 and part of 2022). Results: The findings indicate that higher testing rates, coupled with more stringent containment and public health measures, had a significant negative impact on the death rate in both pre-vaccination and post-vaccination models. The data from the post-vaccination model demonstrate that higher vaccination rates were associated with significant decreases in fatalities. Additionally, our research indicates that countries with healthcare systems characterized by high and medium levels of decommodification experienced lower mortality rates than those with healthcare systems involving low decommodification. Conclusions: The results of this study indicate that stronger public health infrastructure and more inclusive social protections have mitigated the severity of the pandemic's adverse health impacts, more so than emergency containment measures and social restrictions.

• The Journal of Asian Finance, Economics and Business
• /
• 제7권7호
• /
• pp.27-34
• /
• 2020

The study investigates herding behavior in cryptocurrencies in different situations. This study employs daily returns of major cryptocurrencies listed in CCI30 index and sub-major cryptocurrencies and major stock returns listed in Dow-Jones Industrial Average Index, from 2015 to 2018. Quantile regression method is employed to test the herding effect in market asymmetries, inter-dependency and intra-dependency cases. Findings confirm the presence of herding in cryptocurrency in upper quantiles in bullish and high volatility periods because of overexcitement among investors, which lead to high volume trading. Major cryptocurrencies cause herding in sub-major cryptocurrencies, but it is a unidirectional relation. However, no intra-dependency effect among cryptocurrencies and equity market is observed. Results indicate that in the CKK model herding exists at upper quantile in market that may be due when the market is moving fast, continuously trading, and bullish trend are prevailing. Further analysis confirms this narrative as, at upper quantile, the beta of bullish regime is negative and significant, meaning the main source of market herding is a bullish trend in investment, which increases market turbulence and gives investors opportunity to herd. Also, we found that herding in cryptocurrencies exits in high volatility periods, but this herding mostly depends on market activity, not market movement.