home · github

Feature Importance

Parameter importance via GP kernel analysis
About this plot

A bar chart ranking parameters by their influence on the selected outcome. Importance is derived from Sobol sensitivity analysis on the GP model.

Each bar shows two components: the first-order effect (how much variance the parameter explains on its own) and the total-order effect (including interactions with other parameters). A large gap between them means the parameter's effect depends on the values of other parameters.

These importances are also embedded in the sliders of the Ax Explorer.

Lengthscale mode: Bars show 1 / lengthscale — shorter kernel lengthscales mean the model is more sensitive to that parameter. Fast but not range-aware.

Sobol’ mode: Bars show variance decomposition via Saltelli’s estimator on the GP posterior mean. Solid blue = first-order effect (Si), light blue = interaction with other parameters (STi − Si). Range-aware and detects interactions, but requires ~500×(d+2) GP evaluations.