Explaining predictive models using Shapley values and non-parametric vine copulas
From MaRDI portal
Publication:2236381
Abstract: The original development of Shapley values for prediction explanation relied on the assumption that the features being described were independent. If the features in reality are dependent this may lead to incorrect explanations. Hence, there have recently been attempts of appropriately modelling/estimating the dependence between the features. Although the proposed methods clearly outperform the traditional approach assuming independence, they have their weaknesses. In this paper we propose two new approaches for modelling the dependence between the features. Both approaches are based on vine copulas, which are flexible tools for modelling multivariate non-Gaussian distributions able to characterise a wide range of complex dependencies. The performance of the proposed methods is evaluated on simulated data sets and a real data set. The experiments demonstrate that the vine copula approaches give more accurate approximations to the true Shapley values than its competitors.
Recommendations
- Explaining individual predictions when features are dependent: more accurate approximations to Shapley values
- Nonparametric estimation of simplified vine copula models: comparison of methods
- Predictive assessment of copula models
- Prediction based on conditional distributions of vine copulas
- scientific article; zbMATH DE number 5489275
- Predictive inference for bivariate data: combining nonparametric predictive inference for marginals with an estimated copula
- Vine copula models with GLM and sparsity
- Estimating non-simplified vine copulas using penalized splines
- An empirical analysis of multivariate copula models
Cites work
- Adaptive pointwise estimation of conditional density function
- An efficient explanation of individual classifications using game theory
- Analysis of regression in game theory approach
- Analyzing dependent data with vine copulas. A practical guide with R
- Comorbidity of chronic diseases in the elderly: patterns identified by a copula design for mixed responses
- Conditional density estimation using the local Gaussian correlation
- Converting high-dimensional regression to high-dimensional conditional density estimation
- Distribution-free continuous Bayesian belief nets
- Estimation of conditional densities and sensitivity measures in nonlinear dynamical systems
- Evading the curse of dimensionality in nonparametric density estimation with simplified vine copulas
- Explaining individual predictions when features are dependent: more accurate approximations to Shapley values
- Fast kernel conditional density estimation: a dual-tree Monte Carlo approach
- scientific article; zbMATH DE number 3738723 (Why is no real title available?)
- scientific article; zbMATH DE number 3078997 (Why is no real title available?)
- Note on the multivariate Burr's distribution
- On nonlinear, stochastic dynamics in economic and financial time series
- On Shapley Value for Measuring Importance of Dependent Inputs
- Pair copula constructions for multivariate discrete data
- Pair-copula constructions of multiple dependence
- Prediction based on conditional distributions of vine copulas
- Probability density decomposition for conditionally dependent random variables modeled by vines
- Remarks on a Multivariate Transformation
- Remarks on Some Nonparametric Estimates of a Density Function
- Sampling, conditionalizing, counting, merging, searching regular vines
- Shapley Effects for Global Sensitivity Analysis: Theory and Computation
- Structure learning in Bayesian networks using regular vines
- Uncertainty Analysis with High Dimensional Dependence Modelling
- Vines -- a new graphical model for dependent random variables.
Cited in
(6)- Interpretable concept-based classification with Shapley values
- Explaining individual predictions when features are dependent: more accurate approximations to Shapley values
- scientific article; zbMATH DE number 7626724 (Why is no real title available?)
- Robust DC optimal power flow with modeling of solar power supply uncertainty via R-vine copulas
- A comparative study of methods for estimating model-agnostic Shapley value explanations
- Shapley Curves: A Smoothing Perspective
This page was built for publication: Explaining predictive models using Shapley values and non-parametric vine copulas
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2236381)
