DATASET DESIGN SPACE FILLER¶
Identifies sparse regions in an existing design dataset and fills them with new candidate points to improve design-space coverage. Given a set of input columns, it generates a specified number of new input combinations using space-filling sampling, with optional normalization. Use this worker to augment a DOE dataset before surrogate model training or further exploration.
When to use¶
Classification: process.
Tagged: design_space_filler, doe_augmentation, minmax, normalization, sampling, space_filling, sparse_region, znorm.
Inputs¶
| Label | ID | Type | Default | Required | Description |
|---|---|---|---|---|---|
| Dataset | dataset_1 | dataset | — | Input dataset (tabular) to be checked for sparseness; must contain the columns nominated as design inputs — leave unconnected only if inputs are supplied programmatically. | |
| Inputs | inputs | scalar | — | Column names from dataset_1 that define the design-space axes to be analysed and filled; select one or more columns from the dataset picker. | |
| Sampling Size | sampling_size | text | 1000 | Number of candidate points generated internally during the space-filling search; default 1000 — increase for higher-dimensional or denser coverage requirements. | |
| Number of New Inputs | num_new_inputs | text | 11 | Number of new design points to return in the filled dataset; default 11 — set higher when a larger augmentation of the original dataset is needed. | |
| Normalize Data | normalize | string | no | Whether to normalise the input columns before computing distances for sparseness detection; options are ‘yes’ or ‘no’ (default) — enable when input columns have very different value ranges. | |
| Normalization Type | norm_type | string | znorm | Normalisation method applied when normalize=’yes’; ‘znorm’ (default) performs z-score standardisation, ‘minmax’ scales each column to [0, 1]. |
Outputs¶
| Label | ID | Type | Description |
|---|---|---|---|
| Filled Design Space | dataset_filled_design_space | dataset | Tabular dataset containing the newly generated design points (num_new_inputs rows, same input columns as selected) that fill the sparse regions of the original design space. |
Disciplines¶
- ai_ml.preprocessing
- data.dataset.transform
- design_exploration.doe
Runnable example¶
A runnable example is registered for this worker. Open the example workflow on the d3VIEW canvas: /api/workflow/example?id=dataset_design_space_filler
Auto-generated from transformation schema. Worker id: dataset_design_space_filler. Schema hash: 22cc6b9a25f3. Hand-curated docs in workerexamples/ override this page when present.