Article
Mathematics
Tao Hu, Baosheng Liang
Summary: Motivated by the relative loss estimator of the median, a new class of estimators for linear quantile models is proposed using a general relative loss function. The proposed estimator is shown to have smaller variance and be more efficient than traditional linear quantile estimator. Simulation studies and application in a prostate cancer study demonstrate good performance of the proposed method.
Article
Engineering, Marine
Yingguang Wang
Summary: This paper introduces a novel transformation KDE method for calculating sea state parameter distribution tails and extrapolation. The method uses the Box-Cox formulation to transform the original dataset, making it easier to estimate the density using the ordinary KDE method. The results show that this method is more accurate compared to traditional approaches.
Article
Mathematics
Ekaterina Morozova, Vladimir Panov
Summary: This paper discusses extreme value analysis for triangular arrays in mixture models where parameters vary with increasing observations. It introduces a new model for studying maximal stock returns values and demonstrates its effectiveness with numerical examples.
Article
Engineering, Environmental
Helton Saulo, Roberto Vila, Veronica L. Bittencourt, Jeremias Leao, Victor Leiva, George Christakos
Summary: Extreme-value distributions are essential for modeling weather and air pollution events. We introduce a quantile extreme-value Birnbaum-Saunders distribution and its corresponding quantile regression model, and validate the effectiveness of the method through simulation and analysis of real air pollution data.
STOCHASTIC ENVIRONMENTAL RESEARCH AND RISK ASSESSMENT
(2023)
Article
Multidisciplinary Sciences
Ricardo Ramirez-Aldana, Lizbeth Naranjo
Summary: The study demonstrates the importance of using correction terms in linear mixed models to obtain more accurate predictions, especially when considering log-normal distribution. Through simulations and real data analysis, it is proven the significance of correction terms in improving prediction accuracy.
Article
Biology
P. McCullagh, M. F. Tresoldi
Summary: Quantile matching is a method that maps observed response values to quantiles of a target distribution, while a profile likelihood-based criterion is developed for comparing different target distributions in a linear model framework.
Article
Statistics & Probability
Laurens de Haan, Chen Zhou
Summary: This article develops a bootstrap analogue of the asymptotic expansion of the tail quantile process in extreme value theory and applies it to construct confidence intervals for estimators of the extreme value index. It shows the bootstrap consistency of the confidence intervals for the peaks-over-threshold method, but not for the block maxima method. Simulations demonstrate that the sample variance of bootstrapped estimates can be a good approximation for the asymptotic variance of the original estimator.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2022)
Article
Environmental Studies
Bikramaditya Ghosh, Linh Pham, Mariya Gubareva, Tamara Teplova
Summary: This study empirically examines the connection between energy transition metals and global economic and financial sentiment. The findings suggest that the energy transition metal markets impact global sentiment, especially during stressed periods. Furthermore, the risk spillovers between clean energy metals and sentiment indices increase substantially in extreme quantiles, and there is asymmetry in spillovers over time and across quantiles.
Article
Statistics & Probability
Takuma Yoshida
Summary: Extremal quantile regression is explored in this study based on models with varying coefficients. Quantile regression at the tail is important in fields such as finance, meteorology, and environment. Linear models may be unrealistic for some data at the tail, necessitating the use of more flexible models.
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS
(2021)
Article
Statistics & Probability
Martin Bladt, Hansjorg Albrecher, Jan Beirlant
Summary: In this study, we focus on estimating the extreme value index and extreme quantiles for heavy-tailed data that are right-censored. By applying a trimming procedure to the state-of-the-art estimators, new kernel type estimators are derived, with one of them showing competitive performance in extensive simulations. An adaptive selection method is proposed to minimize asymptotic mean squared error in estimation, with an illustration provided using simulated and real-world MTPL insurance data.
ELECTRONIC JOURNAL OF STATISTICS
(2021)
Article
Statistics & Probability
Stephane Girard, Gilles Stupfler, Antoine Usseglio-Carleve
Summary: Expectiles are a least squares analogue of quantiles that have been extensively researched in the fields of actuarial and financial risk assessment over the past decade. Recent papers have focused on the estimation of unconditional extreme expectiles using heavy-tailed observations. The general theory proposed in this study aims to estimate extreme conditional expectiles in heteroscedastic regression models with heavy-tailed noise, with a focus on dealing with high-dimensional covariates. The approach is demonstrated through various examples, including linear models and time series models like ARMA and GARCH.
ANNALS OF STATISTICS
(2021)
Article
Computer Science, Artificial Intelligence
Yafen Ye, Yuanhai Shao, Chunna Li, Xiangyu Hua, Yanru Guo
Summary: The paper introduces an online support vector quantile regression method, Online-SVQR, for dynamic time series with heavy-tailed noise. By using an incremental learning algorithm to update coefficients, Online-SVQR reflects dynamic information and outperforms traditional epsilon-support vector quantile regression in terms of sample selection ability and training speed.
APPLIED SOFT COMPUTING
(2021)
Article
Business, Finance
Jiahao Zhang, Yifeng Zhang, Yu Wei, Zhuo Wang
Summary: This study comprehensively evaluates the impact and connectedness between three important fossil energy markets and four pivotal uncertainties under different market conditions. The empirical findings suggest that the linkages between fossil energy markets and uncertainties are more pronounced under extreme market conditions, and conventional mean-based methods tend to underestimate these linkages.
INTERNATIONAL REVIEW OF ECONOMICS & FINANCE
(2024)
Article
Statistics & Probability
Pallab Kumar Ghosh
Summary: This study introduces a semi-parametric estimation method to analyze the impact of changes in explanatory variables on the unconditional quantile of outcome variables. The results indicate that declining unionization explains a significant portion of the decrease in the 50/10 wage gap in the USA.
JOURNAL OF APPLIED STATISTICS
(2021)
Article
Statistics & Probability
Torsten Hothorn, Achim Zeileis
Summary: This article discusses regression models for supervised learning problems with continuous response, suggesting a more general understanding of regression models as models for conditional distributions. Quantile regression forests are highlighted among algorithms estimating conditional distributions. A novel approach based on a parametric family of distributions characterized by their transformation function is proposed, along with a dedicated transformation tree algorithm for detecting distributional changes. Prediction intervals and inference procedures are provided by the resulting predictive distributions, making them fully parametric yet very general.
JOURNAL OF COMPUTATIONAL AND GRAPHICAL STATISTICS
(2021)
Article
Biology
Tonghui Yu, Liming Xiang, Huixia Judy Wang
Summary: In this paper, a quantile regression for survival data with covariates subject to detection limits is proposed, offering a more flexible tool for modeling the distribution of survival outcomes. The new method utilizes a novel multiple imputation approach to avoid stringent parametric restrictions on censored covariates, demonstrating satisfactory performance in simulation results and application to real data.
Article
Statistics & Probability
Xinyi Li, Li Wang, Huixia Judy Wang
Summary: This article discusses high-dimensional image-on-scalar regression and proposes a flexible partially linear spatially varying coefficient model to investigate the spatial heterogeneity of covariate effects on imaging responses. The spatially varying coefficient functions are approximated via bivariate spline functions over triangulation to address the challenges of spatial smoothing over the imaging response's complex domain. The method can simultaneously perform sparse learning and model structure identification in the presence of ultrahigh-dimensional covariates, accurately and efficiently identifying zero, nonzero constant, and spatially varying components.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2021)
Article
Statistics & Probability
Yingying Hu, Huixia Judy Wang, Xuming He, Jianhua Guo
Summary: This paper introduces a Bayesian joint-quantile regression method that utilizes a linear approximation of quantile coefficients to borrow information across tail quantiles. A new Bayesian estimator for high quantiles is proposed, using a delayed rejection and adaptive Metropolis and Gibbs algorithm, which is demonstrated to be more stable and efficient than conventional methods through numerical studies.
COMPUTATIONAL STATISTICS
(2021)
Article
Computer Science, Interdisciplinary Applications
Zhikun Gao, Yanlin Tang, Huixia Judy Wang, Guangying K. Wu, Jeff Lin
COMPUTATIONAL STATISTICS & DATA ANALYSIS
(2020)
Article
Statistics & Probability
Yingying Zhang, Huixia Judy Wang, Zhongyi Zhu
Summary: This article introduces a more flexible single-index threshold model in the quantile regression setup, dividing the sample based on a linear combination of predictors, and proposes a new estimator. By smoothing the threshold, Gaussian approximation for statistical inference is enabled, allowing characterization of the limiting distribution of the quantile process.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2022)
Article
Biology
Wu Wang, Ying Sun, Huixia Judy Wang
Summary: This paper proposes a functional partially linear regression model to accommodate the heterogeneous relationship between a scalar response and functional covariates. The model is motivated by a study on salinity tolerance of barley families, and it shows higher accuracy in recovering latent groups and estimating functional coefficients compared to existing methods.
Article
Environmental Sciences
Junho Lee, Ying Sun, Huixia Judy Wang
Summary: Spatial cluster detection using a threshold quantile regression model can capture spatial heterogeneity and heteroscedasticity, making it suitable for analyzing spatial clustering phenomena in air quality data. In practical applications, this method can reveal the associations between air pollution and health-related factors in different spatial regions.
Article
Geosciences, Multidisciplinary
Gaurav Agarwal, Ying Sun, Huixia J. Wang
Summary: This paper introduces a copula based multiple indicator kriging model for the analysis of non-Gaussian spatial data, which provides the entire predictive distribution function and demonstrates better predictive performance than traditional methods.
SPATIAL STATISTICS
(2021)
Article
Statistics & Probability
Xiang Peng, Huixia Judy Wang
Summary: The primary goal of subgroup analysis is to identify subgroups of subjects with differential treatment effects. Existing methods may be ineffective when the two distributions differ in scales or in the upper or lower tails. In this study, we propose a new generalized quantile tree method for subgroup identification that is robust and accurate. The practical value of the method is demonstrated through the analysis of an AIDS clinical trial data.
JOURNAL OF COMPUTATIONAL AND GRAPHICAL STATISTICS
(2022)
Article
Statistics & Probability
Wen Xu, Huixia Judy Wang, Deyuan Li
Summary: Quantifying and predicting rare events with significant societal effects is important. This paper proposes a new semiparametric approach based on the tail single-index model to achieve a better balance between model flexibility and parsimony. The proposed method involves three steps and demonstrates its asymptotic properties.
Article
Statistics & Probability
Yanlin Tang, Yinfeng Wang, Huixia Judy Wang, Qing Pan
Summary: Analyzing the tail quantiles of a response distribution is sometimes more important than analyzing the mean in biomarker studies. In this study, a new and simple testing procedure is developed to detect the effects of biomarkers in a high-dimensional quantile regression. Numerical studies show that the proposed method has adequate control of error rate and competitive power.
Article
Engineering, Civil
Tao Gao, Yifei Xu, Huixia Judy Wang, Qiaohong Sun, Lian Xie, Fuqiang Cao
Summary: This study investigates the joint effects of natural climate variability on seasonal precipitation extremes in China from 1961 to 2017. The results indicate that individual climate variability modes have weaker impacts on seasonal extremes compared to mean rainfall. However, the combined effects of multiple large-scale modes are more likely to influence the upper tail of the precipitation distribution in specific seasons. The study also finds significant positive responses of maximum 1-day and 5-day precipitation to the El Nino-Southern Oscillation (ENSO) and Atlantic Multidecadal Oscillation (AMO), and these combined effects are 10 times greater than on mean rainfall.
WATER RESOURCES MANAGEMENT
(2022)
Article
Statistics & Probability
Yinfeng Wang, Huixia Judy Wang, Yanlin Tang
Summary: Motivated by a genome-wide association study, a new robust test for longitudinal data is developed to detect the effects of biomarkers in high-dimensional quantile regression. The test is based on conditional quantile regression and takes advantage of the feature of longitudinal data. Simulation studies demonstrate that the proposed test controls the error rate well and provides competitive power. The method is applied to glomerular filtration rate data to test the overall significance of candidate single-nucleotide polymorphisms associated with Type 1 diabetes.
Article
Statistics & Probability
Xiang Peng, Huixia Judy Wang
Summary: Quantiles and expected shortfalls are commonly used risk measures in financial risk management. In this project, our goal is to develop a stable and practical inference method for the conditional expected shortfall by jointly modeling conditional quantile and expected shortfall. We propose a two-step estimation procedure to reduce computational effort and develop a score-type inference method for hypothesis testing and confidence interval construction, which shows advantages over existing approaches in simulations and empirical studies.
Article
Statistics & Probability
Fengyang He, Huixia Judy Wang, Tiejun Tong