Article
Environmental Sciences
Shailza Sharma, P. P. Mujumdar
Summary: This study investigates the suitability of parametric multivariate extreme value models to correctly represent and estimate the dependence structure of concurrent extremes. The results demonstrate the ability of parametric multivariate models to characterize the complex dependence structure of concurrent extremes.
WATER RESOURCES RESEARCH
(2022)
Article
Statistics & Probability
Clement Chevalier, Olivia Martius, David Ginsbourger
Summary: Modeling the joint distribution of extreme events at multiple locations using max-stable models is a challenging task, but can be achieved by warping weather stations in a latent space of higher dimension. Two methods are proposed to define target dissimilarity matrices, allowing for the capturing of complex spatial dependences of spatial extreme precipitations and reliable extrapolation of functionals such as extremal coefficients. Supplementary materials for this study can be found online.
JOURNAL OF COMPUTATIONAL AND GRAPHICAL STATISTICS
(2021)
Review
Mathematics
Natalia Markovich, Marijus Vaiciulis
Summary: This paper summarizes recent research results on the evolution of random networks and related extreme value statistics, which are of great interest due to their numerous applications. The focus is on the statistical methodology rather than the structure of random networks. The problems arising in evolving networks, particularly due to the heavy-tailed nature of node indices, are discussed. Topics such as tail and extremal indices, preferential and clustering attachments, community detection, stationarity and dependence of graphs, information spreading, and finding influential leading nodes and communities are surveyed. The paper aims to propose possible solutions to unsolved problems and provides a comprehensive review of estimators for tail and extremal indices on random graphs.
Article
Construction & Building Technology
Xun Liu, Weidong Zhuo, Jie Yang
Summary: This paper proposes a new method that combines complexity pursuit (CP) and extreme value theory (EVT) for bridge damage detection. The method shows good detection accuracy and robustness even under environmental and operational variations.
Article
Engineering, Marine
Ed Mackay, Guillaume de Hauteclocque, Erik Vanem, Philip Jonathan
Summary: This study investigates the impact of serial correlation on estimates of environmental extremes in offshore engineering, finding that neglecting serial correlation leads to overestimation of extreme event sizes. A new definition of a sub-asymptotic extremal index is introduced to quantify the effect of neglecting serial correlation, showing that considering serial correlation can reduce over-conservatism. The size of bias in estimates is related to storm event shapes and peak distribution tails, with longer tails leading to larger biases when serial correlation is neglected.
Article
Agricultural Economics & Policy
Luis Fernando Melo-Velandia, Camilo Andres Orozco-Vanegas, Daniel Parra-Amado
Summary: The study found that the prices of perishable foods are more easily affected by extreme weather conditions, with a more significant relationship between dry seasons and perishable food prices. The risk of perishable food price increases gradually with the increase in drought levels. Additionally, changes in the US dollar-Colombian peso exchange rate and fuel prices also impact food prices.
AGRICULTURAL ECONOMICS
(2022)
Article
Engineering, Marine
Rob Shooter, Emma Ross, Agustinus Ribal, Ian R. Young, Philip Jonathan
Summary: The joint extremal spatial dependence of wind speed and significant wave height in the North East Atlantic is quantified using satellite scatterometer and hindcast observations. A multivariate spatial conditional extremes (MSCE) model is applied to analyze the data. The results show that the extremal spatial dependence for wind speed and significant wave height decays over approximately 600-800 km when conditioned on extreme wind speed.
Article
Meteorology & Atmospheric Sciences
Abdelaziz Chaqdid, Alexandre Tuel, Abdelouahad El Fatimy, Nabil El Mocayd
Summary: The history of Morocco is marked by tragic natural disasters caused by floods, resulting in numerous casualties and significant material losses. Extreme precipitation is a major driver of these floods, and understanding its spatial characteristics is crucial for predicting and mitigating the risks they bring. However, the physical drivers of extreme precipitation events in Morocco are not well-known. To address this knowledge gap, researchers used clustering to divide Morocco into regions that exhibit spatial consistency in terms of extreme precipitation. By analyzing atmospheric circulation anomalies during selected extreme precipitation events, they found that Morocco can be divided into five spatially coherent regions, each with different drivers of extreme precipitation.
WEATHER AND CLIMATE EXTREMES
(2023)
Article
Engineering, Electrical & Electronic
Niloofar Mehrnia, Sinem Coleri
Summary: This paper proposes a methodology based on Extreme Value Theory (EVT) for modeling extreme events of a non-stationary wireless channel in the ultra-reliable regime of operation. The approach effectively analyzes extreme events in the channel data sequence and chooses the best model with minimum complexity.
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
(2021)
Article
Mathematics, Interdisciplinary Applications
Graeme Auld, Ioannis Papastathopoulos
Summary: This study examines the distribution of extreme values in non-stationary but identically distributed sequences of random variables, revealing that extremal clustering affects the limiting distribution of appropriately normalized sample maxima. By introducing a new representation, the authors derived the asymptotic distribution for the time between consecutive extreme observations and proposed estimators for measures of extremal clustering. The results are particularly applied to random sequences with periodic dependence structure.
Article
Ergonomics
Pranab Kar, Suvin P. Venthuruthiyil, Mallikarjuna Chunchu
Summary: This study investigates the impact of evasive actions on the sideswipe crash risk of powered two-wheelers (PTWs) on multilane rural highways using Extreme Value Theory. The findings suggest that PTWs experience significant sideswipe crash risk on four-lane and six-lane highways, and the risk increases with the intensity of braking and steering actions. The study emphasizes the importance of incorporating evasive actions in crash risk estimation and developing non-stationary models for more accurate crash frequency estimates.
ACCIDENT ANALYSIS AND PREVENTION
(2023)
Article
Physics, Multidisciplinary
Lior Zarfaty, Eli Barkai, David A. Kessler
Summary: This paper discusses the classic problem of extreme value statistics, showing that the distribution of maxima converges to one of three limiting forms through the Fisher-Tippett-Gnedenko theorem. Utilizing the Gumbel limit allows for accurate approximation of the extreme value distribution, with parameters represented as power series and the underlying distribution transformed. Functional corrections to the Gumbel limit are considered, obtainable through Taylor expansion, which also helps characterize extreme value statistics in cases where the underlying distribution is unknown.
JOURNAL OF PHYSICS A-MATHEMATICAL AND THEORETICAL
(2021)
Article
Engineering, Mechanical
Xin Li, Shaopeng Li, JingYang Li, Yi Su
Summary: This study conducted a pressure measurement test on a rigid segment model under a downburst-like wind to investigate the characteristics of downburst-like wind fields and wind loads. The results showed that the downburst-like wind, wind loads, and evolution power spectrum density (EPSD) exhibited time evolution features, transient effects, and nonstationary time-varying characteristics. The study also estimated the time-varying extreme values of nonstationary random processes and provided corresponding peak factors based on the generalized extreme value (GEV) distribution theory. The fitting accuracy of the GEV distribution was found to be good, and the calculated peak factors of wind pressure and aerodynamic lift were higher than those of wind speed.
PROBABILISTIC ENGINEERING MECHANICS
(2022)
Article
Public, Environmental & Occupational Health
Faizan Nazir, Yasir Ali, Anshuman Sharma, Zuduo Zheng, Md Mazharul Haque
Summary: This study investigates the effects of traditional and connected environments on car-following crash risk using traffic conflict techniques. The results show that the connected environment significantly reduces car-following crash risk.
ANALYTIC METHODS IN ACCIDENT RESEARCH
(2023)
Article
Geosciences, Multidisciplinary
Chi-Hsiang Wang
Summary: This paper presents a statistical method that utilizes maximum recorded values from multiple observational sites to obtain hazard models with reduced bias and variance at high return levels. The method is applied to extreme wind gust data from South Australia and results in significant reduction in bias and variance. The findings suggest that the specified design wind speeds may be conservative for South Australia.
Article
Statistics & Probability
M. Ferreira, H. Ferreira
THEORY OF PROBABILITY AND ITS APPLICATIONS
(2020)
Article
Statistics & Probability
Helena Ferreira, Marta Ferreira
Summary: The paper proposes a new smoothness coefficient to evaluate the degree of smoothness/oscillation in the trajectory of a process, with an intuitive reading and simple estimation. An application to financial series is illustrated.
Article
Statistics & Probability
Helena Ferreira, Ana Paula Martins, Maria da Graca Temido
Summary: This paper discusses a family of models associated with automatic systems that have periodic control, and addresses the issue of filling missing values to enhance signal strength. The relationship between the dependence conditions of the conditional filling method and the extremal index is examined. A consistent estimator for the parameter controlling missing values is proposed, and its properties are analyzed using Markovian sequences.
STATISTICAL PAPERS
(2021)
Article
Statistics & Probability
Helena Ferreira, Marta Ferreira
Summary: The occurrence of successive extreme observations can impact society. Extremal index, a parameter in extreme value theory, is used to evaluate clustering effects of high values. Existing methods for estimating the extremal index depend on two parameters, but we propose a new estimator depending only on one parameter, reducing uncertainty. Simulation results and an application to financial data demonstrate the effectiveness of our approach.
COMMUNICATIONS IN STATISTICS-THEORY AND METHODS
(2023)
Article
Statistics & Probability
Ana Paula Martins, Helena Ferreira, Marta Ferreira
Summary: This paper discusses the risk of atypical phenomena in several areas and proposes a new random field pMAX for modeling extremes. The dependence and pre-asymptotic dependence structure of the field are analyzed, and estimators for the model parameters are obtained.
STATISTICS & PROBABILITY LETTERS
(2022)
Article
Statistics & Probability
Marta Ferreira
Summary: Extreme value theory (EVT) is a set of methods used to infer the risk of various phenomena in different fields. The extremal index is a measure associated with the clustering of extreme values. Estimating the extremal index involves uncertainty in determining the level of high observations and identifying clusters. This study revisits existing estimators, applies automatic choice methods for threshold and clustering parameter, and compares their performance. An application to meteorological data is also presented.
ASTA-ADVANCES IN STATISTICAL ANALYSIS
(2023)
Article
Statistics & Probability
Marta Ferreira
Summary: This study considers the cycles estimator introduced in Ferreira and Ferreira (Ann Inst Henri Poincare Probab Stat 54(2):587-605, 2018) within Extreme Value Theory. A reduced bias estimator based on the Jackknife methodology is presented, along with the application of the bootstrap technique for inference and obtaining confidence intervals. Performance analysis based on simulation indicates that our proposal effectively reduces bias and compares favorably with some well-known methods. Additionally, the methods are applied to real data.
COMPUTATIONAL STATISTICS
(2023)
Article
Statistics & Probability
Helena Ferreira, Marta Ferreira
Summary: The extreme value theory provides specific tools for modeling and predicting extreme phenomena, with risk assessment commonly analyzed through measures for tail dependence and high values clustering. Despite advancements in data collection technology, failures in records can still cause difficulties in statistical inference, especially in the scarce data tail. In this article, a model with a simple and intuitive failures scheme is presented, where each record failure is replaced by the last available record. The extremal behavior of the model with respect to local dependence, high values clustering, and temporal dependence on the tail is studied.
DEPENDENCE MODELING
(2022)
Article
Statistics & Probability
Helena Ferreira, Marta Ferreira
DEPENDENCE MODELING
(2020)
Article
Statistics & Probability
Helena Ferreira, Marta Ferreira
REVSTAT-STATISTICAL JOURNAL
(2020)
