Python package gsa_framework is aimed at providing interface for Global Sensitivity Analysis (GSA). It consists of the following modules:

• sampling
• • random
• • latin hypercube
• • Sobol' quasi-random sequences
• • Saltelli design
• • custom inputs, e.g. obtained from real data / measurements
• sensitivity indices computation
• • Pearson and Spearman correlation coefficients
• • Sobol firt and total order
• • Extended FAST
• • Delta moment-independent indices
• • Feature importances from gradient boosted trees with XGBoost
• models
• • test functions
• • life cycle assessment
• • custom models
• sensitivity analysis that links all of the above for each sensitivity method
• additional components to support reliability of GSA
• • GSA results validation
• • Convergence of sensitivity indices
• • Robustness with bootstrapping

This package is part of the doctoral work of Aleksandra Kim at Paul Scherrer Institute and ETH Zurich.

For detailed API docs, see the versioned API website.