DATASET NORMALIZE (FIT)¶
Standardizes numeric dataset columns using only available (non-missing) values and emits the fitted mean/std parameters as a reusable dataset. Computes population standard deviation (ddof=0); skips id/mid/*_id and non-numeric columns; clamps std<1e-12 to 1.0. Use this on the TRAINING dataset, then feed the ‘parameters’ output into dataset_normalize_apply (to transform a testing dataset with the same parameters) and dataset_normalize_inverse (to recover raw values after prediction).
When to use¶
Tagged: normalization, standardize, zscore, scaling, fit, missing_values, preprocessing.
Inputs¶
| Label | ID | Type | Default | Required | Description |
|---|---|---|---|---|---|
| Dataset | dataset | dataset | — | ✓ | Training dataset to fit and normalize; array-of-row-objects. |
| Columns | columns | text | — | Optional explicit list of columns to normalize; leave empty to auto-select all numeric columns. id/mid/*_id and non-numeric columns are always skipped. | |
| Skip Columns | skip_columns | text | — | Optional extra columns to exclude from normalization (in addition to the automatic id/mid/*_id skip). |
Outputs¶
| Label | ID | Type | Description |
|---|---|---|---|
| Normalized Dataset | dataset | dataset | Input dataset with the selected numeric columns standardized to zero mean / unit std (over available values). |
| Normalization Parameters | parameters | dataset | One row per normalized column with fields column, mean, std. Reuse with dataset_normalize_apply / dataset_normalize_inverse. |
| Status | status | string | Summary of how many columns were normalized vs skipped. |
Disciplines¶
- data.dataset.transform
- data.statistics
Auto-generated from platform schema. Worker id: dataset_normalize_fit. Schema hash: fc4b37b50281. Hand-curated docs in workerexamples/ override this page when present.