The Kendall interaction filter for variable interaction screening in high dimensional classification problems

DOI10.1080/02664763.2022.2031125arXiv2010.06688OpenAlexW4210655835MaRDI QIDQ6134401FDOQ6134401

Author name not available (Why is that?), A. Mkhadri, Karim Oualkacha

Publication date: 25 July 2023

Published in: Journal of Applied Statistics (Search for Journal in Brave)

Abstract: Accounting for important interaction effects can improve prediction of many statistical learning models. Identification of relevant interactions, however, is a challenging issue owing to their ultrahigh-dimensional nature. Interaction screening strategies can alleviate such issues. However, due to heavier tail distribution and complex dependence structure of interaction effects, innovative robust and/or model-free methods for screening interactions are required to better scale analysis of complex and high-throughput data. In this work, we develop a new model-free interaction screening method, termed Kendall Interaction Filter (KIF), for the classification in high-dimensional settings. The KIF method suggests a weighted-sum measure, which compares the overall to the within-cluster Kendall's $au$ of pairs of predictors, to select interactive couples of features. The proposed KIF measure captures relevant interactions for the clusters response-variable, handles continuous, categorical or a mixture of continuous-categorical features, and is invariant under monotonic transformations. We show that the KIF measure enjoys the sure screening property in the high-dimensional setting under mild conditions, without imposing sub-exponential moment assumptions on the features' distributions. We illustrate the favorable behavior of the proposed methodology compared to the methods in the same category using simulation studies, and we conduct real data analyses to demonstrate its utility.

Full work available at URL: https://arxiv.org/abs/2010.06688

zbMATH Keywords

classificationdimension reductionfeatures rankinginteraction screening

Mathematics Subject Classification ID

Statistics (62-XX)

Cites Work

• Sure independence screening in generalized linear models with NP-dimensionality
• Ball Covariance: A Generic Measure of Dependence in Banach Space
• Measuring and testing dependence by correlation of distances
• Title not available (Why is that?)
• Sure Independence Screening for Ultrahigh Dimensional Feature Space
• Feature Screening via Distance Correlation Learning
• Lectures on the Combinatorics of Free Probability
• Nonparametric Independence Screening in Sparse Ultra-High-Dimensional Additive Models
• Factor profiled sure independence screening
• Covariate assisted screening and estimation
• Robust Variable and Interaction Selection for Logistic Regression and General Index Models
• On selecting interacting features from high-dimensional data
• Interaction Screening for Ultrahigh-Dimensional Data
• The Kolmogorov filter for variable screening in high-dimensional binary classification
• Robust rank correlation based screening
• Title not available (Why is that?)
• Finding predictive gene groups from microarray data
• Model-Free Feature Screening for Ultrahigh Dimensional Discriminant Analysis
• The fused Kolmogorov filter: a nonparametric model-free screening method
• Grouped variable screening for ultra-high dimensional data for linear model
• Interaction pursuit in high-dimensional multi-response regression via distance correlation
• A Generic Sure Independence Screening Procedure
• Partition-based ultrahigh-dimensional variable screening
• Innovated interaction screening for high-dimensional nonlinear classification

Cited In (2)

• Interaction identification and clique screening for classification with ultra-high dimensional discrete features
• Interaction screening via Kendall's rank correlation for imbalanced multi-class classification

This page was built for publication: The Kendall interaction filter for variable interaction screening in high dimensional classification problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6134401)