generate_synthetic_data
Generate synthetic data as a CAS table from your column specification. Create mock datasets with configurable columns, distributions, and row counts.
Instructions
Generate a synthetic CAS table from a column specification.
Use this to create realistic mock data on request (e.g. a driver-risk dataset for a demo). Recommended flow: first PROPOSE the column schema to the user in chat and get their agreement, THEN call this tool. The rows are generated in SAS and saved to CAS as a promoted (global) table, immediately usable by the data, charting, AutoML, and scoring tools.
If a table with the requested name already exists, a numbered variant is created automatically (no error). Very large requests are capped to a safe maximum rather than failing.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| seed | No | Random seed for reproducibility (default 12345). | |
| n_rows | No | Number of rows to generate (default 1000). | |
| columns | Yes | List of column specs. Each is an object with ``name`` and ``type`` (one of: id, int, float, category, bool, date) plus type-specific options: - id: sequential zero-padded identifier - int / float: ``min``, ``max``; or ``dist`` "normal" (``mean``, ``std``) or "poisson" (``lambda``); float also accepts ``decimals`` - category: ``levels`` (list) and optional ``weights`` (list) - bool: ``p_true`` (probability of 1; default 0.5) - date: ``start`` and ``end`` as YYYY-MM-DD | |
| server_id | No | CAS server (default cas-shared-default). | cas-shared-default |
| table_name | Yes | Name for the new CAS table. | |
| caslib_name | No | Target caslib (default Public). | Public |
Output Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||