Desirability
Desirability takes the fitted models and a goal for each
measured result, and finds the combination of settings that meets all
goals best. The result is a recommended set of settings, a score D
between 0 (no goal met) and 1 (all goals met), and a report. The
component has a compute button.
Panel: 6 Calibrate. Nickname: Desire. Compute button: yes (runs only when the button is clicked).
Inputs
| Name | Type | Default | Description |
|---|---|---|---|
models |
Generic (list) | RSM Evaluate models or JSON output, one fitted model per response | |
goals |
Text (list) | target=7 in 6..8 w=2; min; min | One goal per model in matching order: min, max, target=7 in 6..8, range=6..8 or none (optional w=2, s=2) |
Optional group: Factor priors (prefer low/high factor levels)
Right-click the component to switch this group on. The inputs appear only while the group is enabled.
| Name | Type | Default | Description |
|---|---|---|---|
factorPriors |
Text (list) | low w=1.5; none; high w=1.5 | Optional prior per factor: ‘none’, ‘low’, ‘high’ or ‘target=x’. Add ‘w=’ for weight. For example prefer a low layer height for bonding |
Optional group: Search bounds / grid resolution
Right-click the component to switch this group on. The inputs appear only while the group is enabled.
| Name | Type | Default | Description |
|---|---|---|---|
lows |
Number (list) | Search minima in actual units, one per factor in model order (otherwise model limits) | |
highs |
Number (list) | Search maxima in actual units, one per factor in model order (otherwise model limits) | |
gridPerDim |
Integer | 15 | Grid resolution per factor, clamped so the total stays at or below 500k points |
feasibleThreshold |
Number | 0.5 | Predicted feasibility below this value is treated as unprintable |
Optional group: Recipe sweep (one recipe per target value)
Right-click the component to switch this group on. The inputs appear only while the group is enabled.
| Name | Type | Default | Description |
|---|---|---|---|
sweepResponse |
Text | Model response name to optimize at each sweepValues target, required for the sweep | |
sweepValues |
Number (list) | 6.5; 7; 7.5; 8 | Target values in sweepResponse units, one optimized recipe per value |
sweepBand |
Number | 0.4 | Acceptance band around each sweep target, plus or minus, in response units |
Optional group: Pareto front (tradeoffs)
Right-click the component to switch this group on. The inputs appear only while the group is enabled.
| Name | Type | Default | Description |
|---|---|---|---|
paretoOn |
Boolean | Collect the non-dominated recipes over the individual goal desirabilities, for Pareto Select. Goal bounds shape the front, weights do not | |
maxFront |
Integer | 200 | Maximum retained tradeoff recipes, preserving each objective’s best recipe even if this cap is exceeded |
Optional group: Excel export input
Right-click the component to switch this group on. The inputs appear only while the group is enabled.
| Name | Type | Description |
|---|---|---|
xlsxPath |
Text | Local .xlsx file or folder to export on Compute with Export enabled, overwriting the target |
Outputs
| Name | Type | Description |
|---|---|---|
settings |
Number (list) | Recommended factor settings in actual units, factor order of the models |
D |
Number | Overall desirability, 0 to 1 |
predicted |
Number (list) | Predicted responses at the recommended settings |
report |
Text | Readable optimization report. Connect a panel |
alternatives |
Number (tree) | Other local best recipes, one branch each: factor values in model order, then overall desirability D |
sweep |
Number (tree) | Sweep recipes, one branch per target: target, factors, D, predictions |
info |
Text | Diagnostics |
interval |
Number (tree) | 95% prediction limits in response units, one branch per response: low, predicted, high (empty without model uncertainty data) |
pareto |
Number (tree) | Tradeoff recipes with Pareto enabled and paretoOn true, one branch each: factors, active-goal desirabilities, D. Wire to Pareto Select front |
objectives |
Text (list) | Names of the active goals in model order, without none goals. The axes of the pareto output. Wire to Pareto Select objectiveNames |
Back to Components.