Use a connector as a data destination

If you use the web application, set a destination from the Dataset destination option when you configure a new synthetic dataset.

Steps

1. Start a new synthetic dataset in one of two ways.
   • From the Synthetic datasets page, click + New synthetic data and select a trained generator.
   • From the Generators page with a trained generator open, click Generate synthetic data.
2. Click Dataset destination.
3. In the modal window, select a destination connector from the Destination drop-down.
4. For Location, follow the steps depending on whether you deliver to a database or cloud storage destination.
   • For databases, select a database and in which schema you want to deliver the synthetic dataset.
   • For cloud storage, click in the Location field and expand the bucket tree structure to select a destination folder.
5. (Optional) Select Overwrite existing table to replace database tables or cloud storage files that have the same name as the tables in your synthetic data.
   📑

   Note: The best use case for overwriting existing tables is when you are delivering multiple times to the same database or cloud bucket.

6. Click Proceed.

Result

Your synthetic dataset now has a destination, and when generated, MOSTLY AI delivers it to the destination.

If you use the Python client, add delivery to your synthetic dataset config dictionary.

To deliver to databases with a schema, use the code snippet below and replace as follows:

• GENERATOR_ID with the generator ID
• DESTINATION_CONNECTOR_ID with the ID of a destination connector
• SCHEMA with the database schema
• TABLE with the table name

g = mostly.generators.get("GENERATOR_ID")
config = {
   "generatorId": "GENERATOR_ID",
   "name": "Databricks US Census",
   "delivery": {
      "destinationConnectorId": "DESTINATION_CONNECTOR_ID",
      "location": "SCHEMA.TABLE"
   }
}
sd = mostly.generate(g, config=config)

To deliver to databases with no schema, use the code snippet below and replace as follows:

• GENERATOR_ID with the generator ID
• DESTINATION_CONNECTOR_ID with the ID of a destination connector
• TABLE with the table name

g = mostly.generate.get("GENERATOR_ID")
config = {
   "generatorId": "GENERATOR_ID",
   "name": "Databricks US Census",
   "delivery": {
      "destinationConnectorId": "DESTINATION_CONNECTOR_ID",
      "location": "TABLE"
   }
}
sd = mostly.generate(g, config=config)

To deliver to cloud buckets, use the code snippet below and replace as follows:

• GENERATOR_ID with the generator ID
• DESTINATION_CONNECTOR_ID with the ID of a destination connector
• path with cloud bucket path to the destination file
• file.csv with the filename and extension (one of .csv, .parquet, or .xslx)

g = mostly.generate.get("GENERATOR_ID")
config = {
   "generatorId": "GENERATOR_ID",
   "name": "Synthetic dataset",
   "delivery": {
      "destinationConnectorId": "DESTINATION_CONNECTOR_ID",
      "location": "/path/file.csv"
   }
}
sd = mostly.generate(g, config=config)