Importance Chart

The Importance Chart ranks parameters by how strongly they influence a selected objective, drawn as a bar chart. You pick the objective and the sensitivity method; each bar is that method's importance score for one parameter, sorted so the most influential parameters are easy to read off.

The scores themselves — and the choice between Spearman, Ridge, tree-based methods, and Sobol — are the subject of the Sensitivity Analysis section.

• Theory and method selection: Sensitivity Analysis
• All objectives at once: the Sensitivity Heatmap shows the same scores for every parameter against every objective in one grid.

Cautions

• Importance scores represent association or predictive contribution in the observed data, not causality. Before drawing conclusions, check whether the same pattern is reproduced with another method.
• Model-based methods, including tree-based methods and Sobol, depend on the quality of the fitted model. If the fit is poor, the displayed ranking and scores may also be unreliable.