Search

A theoretical expression is derived for the mean squared error of a nonparametric estimator of the tail dependence coefficient, depending on a threshold that defines which rank delimits the tails of a distribution. We propose a new method to optimally select this threshold. It combines the theoretical mean squared error of the estimator with a parametric estimation of the copula linking observations in the tails. Using simulations, we compare this semiparametric method with other approaches proposed in the literature, including the plateau-finding algorithm. [ABSTRACT FROM AUTHOR]

We propose a new threshold selection method for nonparametric estimation of the extremal index of stochastic processes. The discrepancy method was proposed as a data-driven smoothing tool for estimation of a probability density function. Now it is modified to select a threshold parameter of an extremal index estimator. A modification of the discrepancy statistic based on the Cramér–von Mises–Smirnov statistic $ \omega ^2 $ ω 2 is calculated by k largest order statistics instead of an entire sample. Its asymptotic distribution as $ k\to \infty $ k → ∞ is proved to coincide with the $ \omega ^2 $ ω 2 -distribution. Its quantiles are used as discrepancy values. The convergence rate of an extremal index estimate coupled with the discrepancy method is derived. The discrepancy method is used as an automatic threshold selection for the intervals and K-gaps estimators. It may be applied to other estimators of the extremal index. The performance of our method is evaluated by simulated and real data examples. [ABSTRACT FROM AUTHOR]

We offer an alternative proof, using the Stein-Chen method, of Bollobás' theorem concerning the distribution of the extreme degrees of a random graph. Our proof also provides a rate of convergence of the extreme degree to its asymptotic distribution. The same method also applies in a more general setting where the probability of every pair of vertices being connected by edges depends on the number of vertices. [ABSTRACT FROM AUTHOR]

This paper derives finite sample results to assess the consistency of Generalized Pareto regression trees introduced by Farkas et al. (Insur. Math. Econ. 98:92–105, 2021) as tools to perform extreme value regression for heavy-tailed distributions. This procedure allows the constitution of classes of observations with similar tail behaviors depending on the value of the covariates, based on a recursive partition of the sample and simple model selection rules. The results we provide are obtained from concentration inequalities, and are valid for a finite sample size. A misspecification bias that arises from the use of a "Peaks over Threshold" approach is also taken into account. Moreover, the derived properties legitimate the pruning strategies, that is the model selection rules, used to select a proper tree that achieves a compromise between simplicity and goodness-of-fit. The methodology is illustrated through a simulation study, and a real data application in insurance for natural disasters. [ABSTRACT FROM AUTHOR]

Let X be a random variable with finite second moment. We investigate the inequality: P { | X - E [ X ] | ≤ Var (X) } ≥ P { | Z | ≤ 1 } , where Z is a standard normal random variable. We prove that this inequality holds for many familiar infinitely divisible continuous distributions including the Laplace, Gumbel, Logistic, Pareto, infinitely divisible Weibull, Log-normal, Student's t and Inverse Gaussian distributions. Numerical results are given to show that the inequality with continuity correction also holds for some infinitely divisible discrete distributions. [ABSTRACT FROM AUTHOR]

This study examines the varying coefficient model in tail index regression. The varying coefficient model is an efficient semiparametric model that avoids the curse of dimensionality when including large covariates in the model. In fact, the varying coefficient model is useful in mean, quantile, and other regressions. The tail index regression is not an exception. However, the varying coefficient model is flexible, but leaner and simpler models are preferred for applications. Therefore, it is important to evaluate whether the estimated coefficient function varies significantly with covariates. If the effect of the non-linearity of the model is weak, the varying coefficient structure is reduced to a simpler model, such as a constant or zero. Accordingly, the hypothesis test for model assessment in the varying coefficient model has been discussed in mean and quantile regression. However, there are no results in tail index regression. In this study, we investigate the asymptotic properties of an estimator and provide a hypothesis testing method for varying coefficient models for tail index regression. [ABSTRACT FROM AUTHOR]

The focus of this article is on studying a partially linear additive model, which is defined using a measurable function ψ : R q → R . The model is given as follows: ψ (Y i) := Y i = Z i ⊤ β + ∑ ℓ = 1 d m ℓ (X ℓ , i) + ε i for 1 ≤ i ≤ n , where Z i = (Z i , 1 , ... , Z i p ) ⊤ and X i = (X 1 , i , ... , X i d ) ⊤ are vectors of explanatory variables, β = (β 1 , ... , β p ) ⊤ is a vector of unknown parameters, m 1 , ... , m d are unknown univariate real functions, and ε 1 , ... , ε n are independent random errors with mean zero, finite variances σε. Additionally, it is assumed that E (ε | X , Z) = 0 almost surely. The main contributions of this article are as follows. First, under certain mild conditions, we establish the asymptotic normality of the non linear additive components of the model. These components are estimated using the marginal integration device with the linear wavelet method. Second, we leverage our main result to construct confidence intervals for the estimated model. [ABSTRACT FROM AUTHOR]

This work focuses on dimension-reduction techniques for modelling conditional extreme values. Specifically, we investigate the idea that extreme values of a response variable can be explained by nonlinear functions derived from linear projections of an input random vector. In this context, the estimation of projection directions is examined, as approached by the extreme partial least squares (EPLS) method—an adaptation of the original partial least squares (PLS) method tailored to the extreme-value framework. Further, a novel interpretation of EPLS directions as maximum likelihood estimators is introduced, utilizing the von Mises–Fisher distribution applied to hyperballs. The dimension reduction process is enhanced through the Bayesian paradigm, enabling the incorporation of prior information into the projection direction estimation. The maximum a posteriori estimator is derived in two specific cases, elucidating it as a regularization or shrinkage of the EPLS estimator. We also establish its asymptotic behavior as the sample size approaches infinity. A simulation data study is conducted in order to assess the practical utility of our proposed method. This clearly demonstrates its effectiveness even in moderate data problems within high-dimensional settings. Furthermore, we provide an illustrative example of the method’s applicability using French farm income data, highlighting its efficacy in real-world scenarios. [ABSTRACT FROM AUTHOR]

In recent years, the direction has turned to non-stationary time series. Here the situation is more complicated: it is often unclear how to set down a meaningful asymptotic for non-stationary processes. For this reason, the theory of locally stationary processes arose, and it is based on infill asymptotics created from non-parametric statistics. The present paper aims to develop a framework for inference of locally stationary functional time series based on the so-called conditional U-statistics introduced by Stute (Ann Probab 19:812–825, 1991), and may be viewed as a generalization of the Nadaraya-Watson regression function estimates. In this paper, we introduce an estimator of the conditional U-statistics operator that takes into account the nonstationary behavior of the data-generating process. We are mainly interested in establishing weak convergence of conditional U-processes in the locally stationary functional mixing data framework. More precisely, we investigate the weak convergence of conditional U-processes when the explicative variable is functional. We treat the weak convergence when the class of functions is bounded or unbounded, satisfying some moment conditions. These results are established under fairly general structural conditions on the classes of functions and the underlying models. The theoretical results established in this paper are (or will be) critical tools for further functional data analysis developments. [ABSTRACT FROM AUTHOR]

Abstract.This study considers excess distribution estimation in iid settings. There are two ways for the estimation; the fitting to the generalized Pareto distribution and the fully non parametric estimation. The fitting estimator is justified by the approximation proven in the extreme value theory; however, the accuracy depends on how extremely large the target is. The non parametric estimator does not need an approximation and has the advantage of wide applicability. This study conducts both theoretical and numerical comparative study on excess distribution estimation. Asymptotic convergence rates of two estimators are obtained, and the mean integrated squared errors are numerically surveyed by simulation study. An illustrative example of Abisko rainfall amount is presented. [ABSTRACT FROM AUTHOR]

This paper develops a new nonparametric estimator of the scalar-on function modal regression that is used to analyse the co-variability between a functional regressor and a scalar output variable. The new estimator inherits the smoothness of the kernel method and the robustness of the quantile regression. We assume that the functional observations are structured as a strong mixing functional time series data and we establish the almost complete consistency (with rate) of the constructed estimator. A discussion highlighting the impact of this new estimator in nonparametric functional data analysis is also given. The usefulness of this new estimator is shown using an artificial data example. [ABSTRACT FROM AUTHOR]

We consider the clustering of extremes for stationary regularly varying random fields over arbitrary growing index sets. We study sufficient assumptions on the index set such that the limit of the point processes of the exceedances above a high threshold exists. Under the so-called anti-clustering condition, the extremal dependence is only local. Thus the index set can have a general form compared to previous literature (Basrak and Planinić in Bernoulli 27(2):1371–1408, 2021; Stehr and Rønn-Nielsen in Extremes 24(4):753–795, 2021). However, we cannot describe the clustering of extreme values in terms of the usual spectral tail measure (Wu and Samorodnitsky in Stochastic Process Appl 130(7):4470–4492, 2020) except for hyperrectangles or index sets in the lattice case. Using the recent extension of the spectral measure for star-shaped equipped space (Segers et al. in Extremes 20:539–566, 2017), the Υ -spectral tail measure provides a natural extension that describes the clustering effect in full generality. [ABSTRACT FROM AUTHOR]

We present necessary and sufficient conditions for the finiteness of moments of a fixed order of generalized order statistics based on an arbitrary life baseline distribution with a finite expectation. The conditions depend on a relation between the moment order and the minimal parameter of the generalized order statistic and its multiplicity. Furthermore, under these conditions we determine sharp upper bounds on the moments of generalized order statistic expressed in the scale units being the respective powers of the mean of the baseline distribution. [ABSTRACT FROM AUTHOR]

We are mainly concerned with the partially linear additive model defined for a measurable function ψ : R q → R , by ψ (Y i) : = Y i = Z i ⊤ β + ∑ l = 1 d m l (X l , i) + ε i for 1 ≤ i ≤ n , where Z i = (Z 1 , i , ... , Z p , i) ⊤ and X i = (X 1 , i , ... , X i , d) ⊤ are vectors of explanatory variables, β = (β 1 , ... , β p) ⊤ is a vector of unknown parameters, m 1 , ... , m d are unknown univariate real functions, and ε 1 , ... , ε n are independent random errors with mean zero, finite variances σ ε and E (ε | X , Z) = 0 a.s. Under some mild conditions, we present a sharp uniform-in-bandwidth limit law for the nonlinear additive components of the model estimated by the marginal integration device with the kernel method. We allow the bandwidth to varying within the complete range for which the estimator is consistent. We provide the almost sure simultaneous asymptotic confidence bands for the regression functions. [ABSTRACT FROM AUTHOR]

We propose and analyze a permutation test of the tail dependence between two random variables whose marginal distributions are assumed to be known. Justifying the test, we show that the proposed test statistics and their permutation distribution converge to the normal distribution. The analysis is motivated by the recent results of DiCiccio and Romano (J Am Stat Assoc 112(519):1211–1220, 2017) on permutation tests for correlation between random variables. We also provide a simulation study of the size and power properties of the test and an application to financial data. [ABSTRACT FROM AUTHOR]

The notion of tail adversarial stability has been proven useful in obtaining limit theorems for tail dependent time series. Its implication and advantage over the classical strong mixing framework has been examined for max-linear processes, but not yet studied for additive linear processes. In this article, we fill this gap by verifying the tail adversarial stability condition for regularly varying additive linear processes. We in addition consider extensions of the result to a stochastic volatility generalization and to a max-linear counterpart. We also address the invariance of tail adversarial stability under monotone transforms. Some implications for limit theorems in statistical context are also discussed. [ABSTRACT FROM AUTHOR]

The sequential tail empirical process is analyzed in a stochastic model allowing for serially dependent observations and heteroscedasticity of extremes in the sense of Einmahl et al. (J. R. Stat. Soc. Ser. B. Stat. Methodol. 78(1), 31–51, 2016). Weighted weak convergence of the sequential tail empirical process is established. As an application, a central limit theorem for an estimator of the extreme value index is proven. [ABSTRACT FROM AUTHOR]

We propose an estimator of the conditional tail moment (CTM) when the data are subject to random censorship. The variable of main interest and the censoring variable both follow a Pareto-type distribution. We establish the asymptotic properties of our estimator and discuss bias-reduction. Then, the CTM is used to estimate, in case of censorship, the premium principle for excess-of-loss reinsurance. The finite sample properties of the proposed estimators are investigated with a simulation study and we illustrate their practical applicability on a dataset of motor third party liability insurance. [ABSTRACT FROM AUTHOR]

Several tests for simultaneously checking homogeneity of means and covariance matrices, based on the classical likelihood ratio and sum-of-squares type test statistics, have been proposed in the literature. Despite their usefulness, they tend to have unsatisfactory size performance for either nonnormal high-dimensional data or strongly spiked eigenvalue models. This article proposes a novel high-dimensional nonparametric test using the maximum-type statistics. The limiting null distribution of the proposed test statistic is derived to provide p-values. It turns out that under appropriate conditions, the test is particularly powerful against sparse alternatives. Since the unknown correlations among the data sometimes pose a great challenge toward the accurate p-value calculation of the test, we further use a parametric bootstrap technique to achieve size accuracy. We also prove that the proposed simulation-based testing procedure is asymptotically valid in terms of size and power even if the dimension of the data is much larger than the sample size. We demonstrate the effectiveness of our methods through extensive simulation studies and a real data application. [ABSTRACT FROM AUTHOR]

The setting of a right-censored random sample subject to contamination is considered. In various fields, expert information is often available and used to overcome the contamination. This paper integrates expert knowledge into the product-limit estimator in two different ways with distinct interpretations. Strong uniform consistency is proved for both cases under certain assumptions on the kind of contamination and the quality of expert information, which sheds light on the techniques and decisions that practitioners may take. The nuances of the techniques are discussed – also with a view towards semi-parametric estimation – and they are illustrated using simulated and real-world insurance data. [ABSTRACT FROM AUTHOR]

We propose new parametrizations for neural networks in order to estimate extreme quantiles in both non-conditional and conditional heavy-tailed settings. All proposed neural network estimators feature a bias correction based on an extension of the usual second-order condition to an arbitrary order. The convergence rate of the uniform error between extreme log-quantiles and their neural network approximation is established. The finite sample performances of the non-conditional neural network estimator are compared to other bias-reduced extreme-value competitors on simulated data. It is shown that our method outperforms them in difficult heavy-tailed situations where other estimators almost all fail. Finally, the conditional neural network estimators are implemented to investigate the behavior of extreme rainfalls as functions of their geographical location in the southern part of France. The source code is available at . [ABSTRACT FROM AUTHOR]

Let α, β be positive real numbers and let X α , β be a Gamma random variable with shape parameter α and scale parameter β. We study infimum values of the function (α , β) ↦ P { X α , β ≤ κ E [ X α , β ] } for any fixed κ > 0 and the function (α , β) ↦ P { | X α , β − E [ X α , β ] | ≤ Var (X α , β) } . We show that inf α , β P { X α , β ≤ E [ X α , β ] } = 1 2 and inf α , β P { | X α , β − E [ X α , β ] | ≤ Var (X α , β) } = P { | Z | ≤ 1 } ≈ 0.6827 , where Z is a standard normal random variable. [ABSTRACT FROM AUTHOR]

In this work, we consider the problem of the approximation of the ruin probability of a classical risk model with large claims using the strong stability method, when the claim distribution is unknown. Since claims are positive variables, we propose a semi-parametric approach to estimate the loss function associated with this distribution. First, a start parametric distribution (Champernowne generalized distribution) is used to transform the initial data. We then apply to the sample resulting from the first step the asymmetric Beta kernel estimator to avoid the problem of boundary effects. Two normalized versions: local (micro-Beta) and global (macro-Beta) are used to avoid the problem of inconsistency of this estimator at boundaries. Simulation studies are performed to support the results. A comparative study between the stability bounds on the ruin probability obtained using the semi-parametric proposed approach and asymmetric non-parametric kernel estimates is carried out to show the performance and the efficiency of the former. [ABSTRACT FROM AUTHOR]

In this paper, the results of El-Adll et al. [Asymptotic prediction for future observations of a random sample of unknown continuous distribution, Complexity 2022 (2022), Art. ID 4073799], are extended to the lower negative tail index distributions. Three distinct estimators of the lower negative tail index are proposed, as well as an asymptotic confidence interval. Moreover, different asymptotic predictive intervals for future observations are constructed for distributions attracted to the lower extreme value distribution with a negative tail index. Furthermore, the asymptotic maximum likelihood estimator (AMLE) of the shape parameter, as well as an asymptotic maximum likelihood predictor (AMLP), are obtained. Finally, extensive simulation studies are conducted to demonstrate the efficiency of the proposed methods. [ABSTRACT FROM AUTHOR]

Extreme value theory motivates estimating extreme upper quantiles of a distribution by selecting some threshold, discarding those observations below the threshold and fitting a generalized Pareto distribution to exceedances above the threshold via maximum likelihood. This sharp cutoff between observations that are used in the parameter estimation and those that are not is at odds with statistical practice for analogous problems such as nonparametric density estimation, in which observations are typically smoothly downweighted as they become more distant from the value at which the density is being estimated. By exploiting the fact that the order statistics of independent and identically distributed observations form a Markov chain, this work shows how one can obtain a natural weighted composite log-likelihood function for fitting generalized Pareto distributions to exceedances over a threshold. A method for producing confidence intervals based on inverting a test statistic calibrated via parametric bootstrapping is proposed. Some theory demonstrates the asymptotic advantages of using weights in the special case when the shape parameter of the limiting generalized Pareto distribution is known to be 0. Methods for extending this approach to observations that are not identically distributed are described and applied to an analysis of daily precipitation data in New York City. Perhaps the most important practical finding is that including weights in the composite log-likelihood function can reduce the sensitivity of estimates to small changes in the threshold. [ABSTRACT FROM AUTHOR]

Weissman extrapolation methodology for estimating extreme quantiles from heavy-tailed distributions is based on two estimators: an order statistic to estimate an intermediate quantile and an estimator of the tail-index. The common practice is to select the same intermediate sequence for both estimators. In this work, we show how an adapted choice of two different intermediate sequences leads to a reduction of the asymptotic bias associated with the resulting refined Weissman estimator. The asymptotic normality of the latter estimator is established and a data-driven method is introduced for the practical selection of the intermediate sequences. Our approach is compared to the Weissman estimator and to six bias reduced estimators of extreme quantiles on a large scale simulation study. It appears that the refined Weissman estimator outperforms its competitors in a wide variety of situations, especially in the challenging high bias cases. Finally, an illustration on an actuarial real data set is provided. [ABSTRACT FROM AUTHOR]

In this paper, we consider the estimation problem for high quantiles of a heavy-tailed distribution from block data when only a few largest values are observed within blocks. We propose estimators for high quantiles and prove that these estimators are asymptotically normal. Furthermore, we employ empirical likelihood method and adjusted empirical likelihood method to constructing the confidence intervals of high quantiles. Through a simulation study we also compare the performance of the normal approximation method and the adjusted empirical likelihood methods in terms of the coverage probability and length of the confidence intervals. [ABSTRACT FROM AUTHOR]

Let Poisson variables X 1 , ⋯ , X m be independent and Y 1 , ⋯ , Y n be also independent such that each X i ∼ P (λ i) , Y j ∼ P (μ j). We prove that if X m : m = max { X 1 , ⋯ , X m } and Y n : n = max { Y 1 , ⋯ , Y n } have the same distribution, then m = n and { λ 1 , ⋯ , λ m } = { μ 1 , ⋯ , μ m }. The weak version in term of minimum statistics is also derived. [ABSTRACT FROM AUTHOR]

We consider the problem of nonparametric estimation of certain conditional models when the explanatory variable is functional. The models studied are the Conditional Expectation (CE), the Conditional Distribution Function (CDF) and the Conditional Probability Density (CPD). The estimators are constructed by combining the local linear method and the k-Nearest Neighbors (kNN) smoothing approach. For each of the three models, we define an optimal estimator associated with the best number of neighbors chosen using two bandwidth selection procedures which are the cross-validation criterion and the bootstrap smoothing rule. The asymptotic properties of these optimal kNN Local Linear Estimators ( k NN- LLE) are established under some standard conditions. The performances of the finite samples of these estimators are then examined and compared through a Monte-carlo study. In addition, an application on the riboflavin content in the yogurt using the Near-infrared curves is carried out to demonstrate the usefulness of the models proposed in the point-wise prediction as well as for the predictive interval. [ABSTRACT FROM AUTHOR]

Motivated by the Extreme Value Analysis 2021 (EVA 2021) data challenge, we propose a method based on statistics and machine learning for the spatial prediction of extreme wildfire frequencies and sizes. This method is tailored to handle large datasets, including missing observations. Our approach relies on a four-stage, bivariate, sparse spatial model for high-dimensional zero-inflated data that we develop using stochastic partial differential equations (SPDE), allowing sparse precision matrices for the latent processes. In Stage 1, the observations are separated in zero/nonzero categories and modeled using a two-layered hierarchical Bayesian sparse spatial model to estimate the probabilities of these two categories. In Stage 2, we first obtain empirical estimates of the spatially-varying mean and variance profiles across the spatial locations for the positive observations and smooth those estimates using fixed rank kriging. This approximate Bayesian inference method is employed to avoid the high computational burden of large spatial data modeling using spatially-varying coefficients. In Stage 3, we further model the standardized log-transformed positive observations from the second stage using a sparse bivariate spatial Gaussian process. The Gaussian distribution assumption for wildfire counts developed in the third stage is computationally effective but erroneous. Thus, in Stage 4, the predicted exceedance probabilities are post-processed using Random Forests. We draw posterior inference for Stages 1 and 3 using Markov chain Monte Carlo (MCMC) sampling. We then create a cross-validation scheme for the artificially generated gaps and compare the EVA 2021 prediction scores of the proposed model to those obtained using some competitors. [ABSTRACT FROM AUTHOR]

In this paper we propose a procedure to estimate the distribution of wildfire frequency and severity using the wildfire data measured by month during 1993–2015. To this end, a spatial quantile autoregressive model (SQAR) is applied to the data with an aid of extreme value theory. Using the proposed method we are able to predict the distributional behavior of the data and identify the hidden structures beyond their mean structures. In addition, abundant interpretations are available with a regression-based model. We provide the estimated results from the wildfire data, including significant explanatory variables and some meaningful interpretations. [ABSTRACT FROM AUTHOR]

This paper details the approach of the team Kohrrelation in the 2021 Extreme Value Analysis data challenge, dealing with the prediction of wildfire counts and sizes over the contiguous US. Our approach uses ideas from extreme-value theory in a machine learning context with theoretically justified loss functions for gradient boosting. We devise a spatial cross-validation scheme and show that in our setting it provides a better proxy for test set performance than naive cross-validation. The predictions are benchmarked against boosting approaches with different loss functions, and perform competitively in terms of the score criterion, finally placing second in the competition ranking. [ABSTRACT FROM AUTHOR]

This paper studies the estimation of extreme conditional quantile for single-index varying-coefficient models (SIVCM). Our contribution is a three-step extreme quantile regression procedure for SIVCM. In the first step, we estimate the intermediate conditional quantile in a quantile regression framework. In the second step, we estimate the extreme value index (EVI) based on the estimated intermediate quantile. In the third step, we extrapolate the intermediate conditional quantile to the extreme tails by adapting the extreme value theory (EVT). We obtain the asymptotation of tail single-index varying-coefficient estimators and demonstrate through simulation study and real data example that the proposed methods enjoy higher accuracy than the conventional quantile regression estimates. [ABSTRACT FROM AUTHOR]