On Robustness of the Explanatory Power of Machine Learning Models: Insights From a New Explainable AI Approach Using Sensitivity Analysis

Abstract Machine learning (ML) is increasingly considered the solution to environmental problems where limited or no physico‐chemical process understanding exists. But in supporting high‐stakes decisions, where the ability to explain possible solutions is key to their acceptability and legitimacy, M...

Full description

Saved in:
Bibliographic Details
Main Authors: Banamali Panigrahi, Saman Razavi, Lorne E. Doig, Blanchard Cordell, Hoshin V. Gupta, Karsten Liber
Format: Article
Language:English
Published: Wiley 2025-03-01
Series:Water Resources Research
Subjects:
Online Access:https://doi.org/10.1029/2024WR037398
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Abstract Machine learning (ML) is increasingly considered the solution to environmental problems where limited or no physico‐chemical process understanding exists. But in supporting high‐stakes decisions, where the ability to explain possible solutions is key to their acceptability and legitimacy, ML can fall short. Here, we develop a method, rooted in formal sensitivity analysis, to uncover the primary drivers behind ML predictions. Unlike many methods for explainable artificial intelligence (XAI), this method (a) accounts for complex multi‐variate distributional properties of data, common in environmental systems, (b) offers a global assessment of the input‐output response surface formed by ML, rather than focusing solely on local regions around existing data points, and (c) is scalable and data‐size independent, ensuring computational efficiency with large data sets. We apply this method to a suite of ML models predicting various water quality variables in a pilot‐scale experimental pit lake. A critical finding is that subtle alterations in the design of some ML models (such as variations in random seed, functional class, hyperparameters, or data splitting) can lead to different interpretations of how outputs depend on inputs. Further, models from different ML families (decision trees, connectionists, or kernels) may focus on different aspects of the information provided by data, despite displaying similar predictive power. Overall, our results underscore the need to assess the explanatory robustness of ML models and advocate for using model ensembles to gain deeper insights into system drivers and improve prediction reliability.
ISSN:0043-1397
1944-7973