Inferring feature importance with uncertainties with application to large genotype data

Abstract

Estimating feature importance, which is the contribution of a prediction or several predictions due to a feature, is an essential aspect of explaining data-based models. Besides explaining the model itself, an equally relevant question is which features are important in the underlying data generating process. We present a Shapley-value-based framework for inferring the importance of individual features, including uncertainty in the estimator. We build upon the recently published model-agnostic feature importance score of SAGE (Shapley additive global importance) and introduce Sub-SAGE. For tree-based models, it has the advantage that it can be estimated without computationally expensive resampling. We argue that for all model types the uncertainties in our Sub-SAGE estimator can be estimated using bootstrapping and demonstrate the approach for tree ensemble methods. The framework is exemplified on synthetic data as well as large genotype data for predicting feature importance with respect to obesity.

SARS-CoV-2 Genetic Variants and Patient Factors Associated with Hospitalization Risk

“…Finally, while we evaluated the overall contribution of genetic variation in both the XGBoost and GLMM models, we have not evaluated the statistical significance of individual amino acid changes. For the XGBoost-SHAP analysis, methods for evaluating statistical significance are an active area of research (50). In the case of GLMMs, we ran separate versions of the model for each of 28 amino acid changes and did not apply multiple-testing correction methods.…”