Saved in:
Bibliographic Details
Main Authors: Bladen, Kelvyn K., Cutler, Adele, Cutler, D. Richard, Moon, Kevin R.
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.08821
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911503632826368
author Bladen, Kelvyn K.
Cutler, Adele
Cutler, D. Richard
Moon, Kevin R.
author_facet Bladen, Kelvyn K.
Cutler, Adele
Cutler, D. Richard
Moon, Kevin R.
contents Global variable importance measures are commonly used to interpret the results of machine learning models. Local variable importance techniques assess how variables contribute to individual observations. Current, popular methods, including LIME and SHAP, typically fail to accurately reflect locally dependent relationships between variables and instead focus on marginal importance values. Additionally, they are not natively adapted for multi-class classification problems. We propose a new model-agnostic method for calculating local variable importance, CLIQUE, that captures locally dependent relationships, provides improvements over permutation-based methods, and can be directly applied to multi-class classification problems. Simulated and real-world examples show that CLIQUE emphasizes locally dependent information, captures interaction behavior beyond what can be evaluated by correlations, and properly reduces bias in regions where variables do not affect the response.
format Preprint
id arxiv_https___arxiv_org_abs_2411_08821
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Conditional Local Importance by Quantile Expectations
Bladen, Kelvyn K.
Cutler, Adele
Cutler, D. Richard
Moon, Kevin R.
Machine Learning
Computation
Global variable importance measures are commonly used to interpret the results of machine learning models. Local variable importance techniques assess how variables contribute to individual observations. Current, popular methods, including LIME and SHAP, typically fail to accurately reflect locally dependent relationships between variables and instead focus on marginal importance values. Additionally, they are not natively adapted for multi-class classification problems. We propose a new model-agnostic method for calculating local variable importance, CLIQUE, that captures locally dependent relationships, provides improvements over permutation-based methods, and can be directly applied to multi-class classification problems. Simulated and real-world examples show that CLIQUE emphasizes locally dependent information, captures interaction behavior beyond what can be evaluated by correlations, and properly reduces bias in regions where variables do not affect the response.
title Conditional Local Importance by Quantile Expectations
topic Machine Learning
Computation
url https://arxiv.org/abs/2411.08821