gsaot: an R package for Optimal Transport-based sensitivity analysis

2025-07-24

AI Summary

This paper addresses the model dependency and low computational efficiency inherent in conventional global sensitivity analysis (GSA). To overcome these limitations, we propose a model-agnostic sensitivity measure framework grounded in optimal transport (OT) theory, implemented as the open-source R package `gsaot`. The package provides a unified interface to state-of-the-art OT solvers—including the network simplex algorithm and the Sinkhorn–Knopp iterative method—enabling post-hoc sensitivity assessment for arbitrary black-box models. It integrates efficient estimation of sensitivity indices, statistical inference capabilities, and interactive visualization tools. Extensive experiments on diverse benchmark models demonstrate the method’s accuracy, robustness to sampling variability, and scalability to high-dimensional inputs. By abstracting away implementation complexity, `gsaot` substantially lowers the technical barrier to applying OT-based GSA. It delivers a lightweight, general-purpose, and fully reproducible R solution for uncertainty quantification in scientific and engineering applications.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Reasoning under Uncertainty: Stochastic Optimization
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

gsaot is an R package for Optimal Transport-based global sensitivity analysis. It provides a simple interface for indices estimation using a variety of state-of-the-art Optimal Transport solvers such as the network simplex and Sinkhorn-Knopp. The package is model-agnostic, allowing analysts to perform the sensitivity analysis as a post-processing step. Moreover, gsaot provides functions for indices and statistics visualization. In this work, we provide an overview of the theoretical grounds, of the implemented algorithms, and show how to use the package in different examples.

Problem

• Develops R package for Optimal Transport sensitivity analysis
• Estimates indices using advanced Optimal Transport solvers
• Enables model-agnostic post-processing and visualization

Innovation

• R package for Optimal Transport sensitivity analysis
• Uses network simplex and Sinkhorn-Knopp solvers
• Model-agnostic post-processing visualization functions