Identify and Define Golden Batch
Single guided flow
Start with visualization, carry your selection forward, then review and lock without switching pages.
Upstream results are stale.
Schema Mapping
Assign semantic roles once; downstream calculations use this shared mapping.
Preparing automatic suggestions…
| Column | 5 sample values | Unit | Role |
|---|
KPI Computation
Compute in progress… KPI results will appear here when the job completes.
| MV_FLOW column | Unit | Price/kg | Currency | Remove material |
|---|
Multi-Criteria Separator
Build a reviewable filter chain without changing the committed pool.
5 sample values: Loading KPI values...
Funnel: —
| Step | Kind | Expression | Enabled | Prefix count | Actions |
|---|
| Surviving batches (sorted) |
|---|
Feature Importance
Choose Y, then compute a held-out permutation-importance ranking from the original Step 1 upload. KPI and Multi-Criteria results are not used; IDs, time/index, STATE, and reference/control fields are excluded.
Method: held-out permutation importance
Choose the variable you want to explain or visualize. The ranking shows which other original-upload variables are most useful for predicting it.
Load a dataset first to visualize batches or upload an ideal batch.
Workflow tip: visualize Y first, select/preview batches, then lock. Locked batches are used as the golden reference for downstream PCA simulation.
1. Choose Y and Visualize
Each click/lasso replaces the previous plot selection
Numeric Y options: 0
Current Y: -
Example: Penicillin concentration (P:g/L)
Quick workflow
1.Choose the Y variable that best reflects output quality or stability.
2.Visualize all batch trajectories to spot stable, tight, or high-performing runs.
3.Click or lasso promising batches, then preview them before locking.
Step 1
Visualize batch trajectories
Choose a Y variable and click Visualize Batch-wise to compare all batches over time.
Ideal Batch Similarity Matrix
Similarity values for each batch and measurementUpload an ideal batch to compare it with the active batches.
Similarity-Based Batch Visualization
Compare all matching batches with the Ideal BatchChoose a category and common Y variable.
Select a Y variable to display batch trajectories.
2. Select Best Performing Batches
Manual list or automatic by Y range
Visualize a Y variable to unlock selection modes.
Use plot selection or choose from the list below. Check boxes for multi-select.
0 batches selected
Filter batches by the final recorded Y value to build a first-pass shortlist.
Using Y = -
3. Look the Trajectory and lock Golden Batches
Finalize selection and continue4. X-Variable Check
Verify the same golden batches across other input variables
Select an X variable to plot it for the currently previewed golden batches.
Selected Best Performing Batches
Use Add Batch to include more batches. Click the x on any selected batch chip to remove it.Selected: 0
What this means
These batches define your Golden Batch baseline. Statistics and similarity analysis are built from this set.
How to update
Use Add Batch to include more, click x on chip to remove, then lock again to save the latest baseline.