Principal component analysis for multivariate extremes
From MaRDI portal
Publication:2044326
Abstract: The first order behavior of multivariate heavy-tailed random vectors above large radial thresholds is ruled by a limit measure in a regular variation framework. For a high dimensional vector, a reasonable assumption is that the support of this measure is concentrated on a lower dimensional subspace, meaning that certain linear combinations of the components are much likelier to be large than others. Identifying this subspace and thus reducing the dimension will facilitate a refined statistical analysis. In this work we apply Principal Component Analysis (PCA) to a re-scaled version of radially thresholded observations. Within the statistical learning framework of empirical risk minimization, our main focus is to analyze the squared reconstruction error for the exceedances over large radial thresholds. We prove that the empirical risk converges to the true risk, uniformly over all projection subspaces. As a consequence, the best projection subspace is shown to converge in probability to the optimal one, in terms of the Hausdorff distance between their intersections with the unit sphere. In addition, if the exceedances are re-scaled to the unit ball, we obtain finite sample uniform guarantees to the reconstruction error pertaining to the estimated projection sub-space. Numerical experiments illustrate the relevance of the proposed framework for practical purposes.
Recommendations
- Principal component analysis in an asymmetric norm
- Sparse representation of multivariate extremes with applications to anomaly detection
- On consistency and sparsity for principal components analysis in high dimensions
- Sparse PCA: optimal rates and adaptive estimation
- High-dimensional principal projections
Cites work
- \(k\)-means clustering of extremes
- Asymptotic Statistics
- Asymptotic Theory for Principal Component Analysis
- Bias correction in multivariate extremes
- Decompositions of dependence for high-dimensional extremes
- Determining the dependence structure of multivariate extremes
- Dimension reduction in multivariate extreme value analysis
- Heavy-Tail Phenomena
- scientific article; zbMATH DE number 1246230 (Why is no real title available?)
- scientific article; zbMATH DE number 4000257 (Why is no real title available?)
- Identifying groups of variables with the potential of being large simultaneously
- Learning Theory
- Likelihood-based inference for max-stable processes
- MAX-stable models for multivariate extremes
- Models for stationary max-stable random fields
- Multivariate generalized Pareto distributions: parametrizations, representations, and properties
- New asymptotic results in principal component analysis
- Nonasymptotic upper bounds for the reconstruction error of PCA
- Nonparametric estimation of the spectral measure of an extreme value distribution.
- On the Eigenspectrum of the Gram Matrix and the Generalization Error of Kernel-PCA
- Probability Inequalities for Sums of Bounded Random Variables
- Random matrix approximation of spectra of integral operators
- Rank-based inference for bivariate extreme-value copulas
- Regular variation for measures on metric spaces
- Regularly varying measures on metric spaces: hidden regular variation and hidden jumps
- Robust bounds in multivariate extremes
- Simulating multivariate extreme value distributions of logistic type
- Sparse representation of multivariate extremes with applications to anomaly detection
- Tail dimension reduction for extreme quantile estimation
- The generalized Pareto process; with a view towards application and simulation
Cited in
(22)- Extreme partial least-squares
- Dimension reduction in multivariate extreme value analysis
- Principal component analysis in an asymmetric norm
- Sparse regular variation
- Total positivity in multivariate extremes
- Tail inverse regression: dimension reduction for prediction of extremes
- Simulating flood event sets using extremal principal components
- Regular variation in Hilbert spaces and principal component analysis for functional extremes
- Concentration bounds for the empirical angular measure with statistical learning applications
- Cross-validation on extreme regions
- Maximum likelihood estimation of elliptical tail
- Information criteria for the number of directions of extremes in high-dimensional data
- Estimation of the number of principal components in high-dimensional multivariate extremes
- On regression in extreme regions
- Identifying regions of concomitant compound precipitation and wind speed extremes over Europe
- Insights into kernel PCA with application to multivariate extremes
- Multivariate generalized Pareto distributions along extreme directions
- A wee exploration of techniques for risk assessments of extreme events. EVA (2023) Conference Data Challenge: Wee Extremes group
- Mathematics, statistics, and geometry of extreme events in high dimensions. Abstracts from the workshop held August 19--23, 2024
- Estimation of the spectral measure from convex combinations of regularly varying random vectors
- Characterizing extremal dependence on a hyperplane
- Multivariate analysis in dam monitoring data with PCA
This page was built for publication: Principal component analysis for multivariate extremes
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2044326)
