Pareto Scatter 2D
The Pareto Scatter 2D chart displays trial values for two selected objective functions as a scatter plot. Its color classification is a 2D projection of the Pareto ranking computed using all objectives, not an independent ranking using only the selected X/Y axes. Changing the axes changes the projection, not the ranks.
Controls
| Item | Description |
|---|---|
| X / Y axis | Use the X and Y dropdowns to select which objective is plotted on each axis. |
| Infeasible trials (constrained studies only) | These are always shown in a distinct color when the study has constraints; there is no control to hide them. |
Operations
Pan, zoom, reset, point hover/click, and Shift-drag region select follow the standard 2D chart controls.
| Item | Description |
|---|---|
| Hover | The point tooltip shows the trial number and objective values. |
| Trial detail | Click a point to open its detail modal. |
How to Read
Dominance is evaluated using all objectives and each objective's minimize or maximize direction. In constrained studies, only feasible trials are compared for this Pareto ranking; infeasible trials are shown separately.
- Pareto front points (red): rank-0, non-dominated trials. No other trial is at least as good in all objectives and strictly better in at least one.
- Non-Pareto points (blue): dominated trials with rank greater than 0. Another trial is at least as good in all objectives and strictly better in at least one.
- Infeasible points: trials that violate constraints, shown in a distinct color.
- Unselected points (greyed): points outside the current selection region.
- Brushing/linking: selections made in other widgets are received and highlighted here.
With three or more objectives, a red point can look dominated in the 2D projection. For example, consider two feasible trials with all three objectives minimized: A = (1, 1, 0) and B = (0, 0, 2). Neither dominates the other in 3D. When the first and second objectives are the axes, B dominates A in the projection, but both are on the full-objective Pareto front (rank 0) in this two-trial example.