> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/sdgm/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/sdgm/_mcp/server.

# How do I generate predictions on new data?

> Run a trained AutoML model after refreshing its graph data

Create a new batch prediction with the trained model. Kumo refreshes the graph data before the prediction unless that data was refreshed recently, allowing the existing model to run on newly available rows.

Retrain the predictive query when the model should learn from the new data rather than only score it. You can run these workflows from the Kumo interface or automate them with the SDK.

See [Batch Predictions](/batch-prediction). If the new data uses a separate graph definition, see [Can I predict with a different graph?](/troubleshooting/how-do-i-make-inference-on-a-different-graph).