XAItest

XAItest identifies features using eXplainable AI (XAI) methods such as SHAP or LIME and enables comparison of these methods with traditional statistical tests.

Key Features:

• XAI-based feature identification: Identifies features using SHAP or LIME.
• Method comparison: Compares XAI methods with traditional statistical tests including t-tests, empirical Bayes, and Fisher's test.
• simThresh system: Enables comparison of feature importance with p-values by incorporating calibrated simulated data.

Scientific Applications:

• Feature discovery: Enhances feature discovery by integrating explainable AI methods with conventional statistical approaches.
• Method benchmarking: Benchmarks XAI methods against traditional statistical tests for feature identification.

Methodology:

XAItest implements SHAP and LIME for feature identification, and includes simThresh for comparing feature importance with p-values using calibrated simulated data.