datarobot-predictions

To generate batch predictions, provide a properly structured CSV file containing the required features for your DataRobot deployment. The Skill validates the CSV structure and scores the inputs against the deployment endpoint, returning prediction outputs for each row.

Yes, you can request per-row prediction explanations using SHAP or XEMP. When calling the DataRobot prediction endpoints, specify max_explanations and related parameters to retrieve top-N contributor explanations for each scored row.

A prediction dataset template is a correctly structured CSV file with sample values matching your deployment's required feature schemas. The Skill generates this template by retrieving deployment feature requirements and producing properly typed input rows.

Yes, the Skill supports real-time scoring for single rows in addition to batch scoring against CSV data. It identifies deployment-required feature schemas and produces properly typed input rows to call the DataRobot prediction endpoints.

Pandas and CSV are listed dependencies for batch scoring operations. You need a valid CSV file containing the deployment's required features, which the Skill validates before calling DataRobot prediction endpoints to return scored outputs.

CSV validation fails when the file structure does not match the deployment's required feature schemas or contains improperly typed rows. The Skill retrieves feature requirements and validates the prediction CSV to ensure inputs are correctly structured before scoring.