Skip to content

On-premise users: click in-app to access the full platform documentation for your version of DataRobot.

Predictions > Batch prediction methods > Prediction monitoring jobs > Monitoring jobs API

Monitoring jobs API

This integration creates a Batch Monitoring API with batchMonitoringJobDefinitions and batchJobs endpoints, allowing you to create monitoring jobs. Monitoring job intake and output settings are configured using the same options as batch prediction jobs. Use the following routes, properties, and examples to create monitoring jobs:

Service health information for external models and monitoring jobs

Service health information such as latency, throughput, and error rate is unavailable for external, agent-monitored deployments or when predictions are uploaded through a prediction monitoring job.

Time series model consideration

Monitoring jobs don't support monitoring predictions made by time series models.

Monitoring job definition and batch job routes

batchMonitoringJobDefinitions endpoints

Access endpoints for performing operations on batch monitoring job definitions:

Operation and endpoint Description
POST /api/v2/batchMonitoringJobDefinitions/ Create a monitoring job definition given a payload.
GET /api/v2/batchMonitoringJobDefinitions/ List all monitoring job definitions.
GET /api/v2/batchMonitoringJobDefinitions/{monitoringJobDefinitionId}/ Retrieve the specified monitoring job definition.
DELETE /api/v2/batchMonitoringJobDefinitions/{monitoringJobDefinitionId}/ Delete the specified monitoring job definition.
PATCH /api/v2/batchMonitoringJobDefinitions/{monitoringJobDefinitionId}/ Update the specified monitoring job definition given a payload.

batchJobs endpoints

Access endpoints for performing operations on batch jobs:

Operation and endpoint Description
POST /api/v2/batchJobs/fromJobDefinition/ Launch (run now) a monitoring job from a monitoringJobDefinition. The payload should contain the monitoringJobDefinitionId.
GET /api/v2/batchJobs/ List the full history of monitoring jobs, including running, aborted, and executed jobs.
GET /api/v2/batchJobs/{monitoringJobId}/ Retrieve a specific monitoring job.
DELETE /api/v2/batchJobs/{monitoringJobId}/ Abort a running monitoring job.

Monitoring job properties

monitoringColumns properties

Define which columns to use for batch monitoring:

Property Type Description
predictionsColumns string (Regression) The column in the data source containing prediction values. You must provide this field and/or actualsValueColumn.
predictionsColumns array (Classification) The columns in the data source containing each prediction class. You must provide this field and/or actualsValueColumn. (Supports a maximum of 1000 items)
associationIdColumn string The column in the data source which contains the association ID for predictions.
actualsValueColumn string The column in the data source which contains actual values. You must provide this field and/or predictionsColumns.
actualsTimestampColumn string The column in the data source which contains the timestamps for actual values.

monitoringOutputSettings properties

Configure the output settings specific to monitoring jobs:

Property Type Description
uniqueRowIdentifierColumns array Columns from the data source that will serve as unique identifiers for each row. These columns are copied to the data destination to associate each monitored status with its corresponding source row. (Supports a maximum of 100 items)
monitoredStatusColumn string The column in the data destination containing the monitoring status for each row.

Note

For general batch job output settings, see the Prediction output settings documentation.

monitoringAggregation properties

To support challengers for external models with large-scale monitoring enabled (meaning that raw data isn't stored in the DataRobot platform), you can report a small sample of raw feature and predictions data; then, you can send the remaining data in aggregate format. Configure the retention settings to indicate that raw data is aggregated by the MLOps library and define how much raw data should be retained for challengers.

Autosampling for large-scale monitoring

To automatically report a small sample of raw data for challenger analysis and accuracy monitoring, you can define the MLOPS_STATS_AGGREGATION_AUTO_SAMPLING_PERCENTAGE when enabling large-scale monitoring for an external model.

Property Type Description
retentionPolicy string The policy definition determines if the retentionValue represents a number of samples or a percentage of the dataset.
enum: ['samples', 'percentage']
retentionValue integer The amount of data to retain, either a percentage of data or the number of samples.

If you define these properties, raw data is aggregated by the MLOps library. This means that the data isn't stored in the DataRobot platform. Stats aggregation only supports feature and prediction data, not actuals data for accuracy monitoring. If you've defined actualsValueColumn or associationIdColumn (which means actuals will be provided later), DataRobot cannot aggregate data.

Preview: Accuracy monitoring with aggregation

Now available for preview, monitoring jobs for external models with aggregation enabled can support accuracy tracking. With this feature enabled, when you configure the retention settings and define the actualsValueColumn for accuracy monitoring with aggregation enabled, you must also define the predictionsColumns and associationIdColumn.

Feature flag OFF by default: Enable Accuracy Aggregation

Monitoring job examples

Regression
{
  "batchJobType": "monitoring",
  "deploymentId": "<deployment_id>",
  "intakeSettings": {
      "type": "jdbc",
      "dataStoreId": "<data_store_id>",
      "credentialId": "<credential_id>",
      "table": "lending_club_regression",
      "schema": "SCORING_CODE_UDF_SCHEMA",
      "catalog": "SANDBOX"
  },
  "outputSettings": {
      "type": "jdbc",
      "dataStoreId": "<data_store_id>",
      "table": "lending_club_regression_out",
      "catalog": "SANDBOX",
      "schema": "SCORING_CODE_UDF_SCHEMA",
      "statementType": "insert",
      "createTableIfNotExists": true,
      "credentialId": "<credential_id>",
      "commitInterval": 10,
      "whereColumns": [],
      "updateColumns": []
  },
  "passthroughColumns": [],
  "monitoringColumns": {
      "predictionsColumns": "PREDICTION",
      "associationIdColumn": "id",
      "actualsValueColumn": "loan_amnt"
  },
  "monitoringOutputSettings": {
     "monitoredStatusColumn": "monitored",
     "uniqueRowIdentifierColumns": ["id"]
  }
  "schedule": {
      "minute": [ 0  ],
      "hour": [ 17   ],
      "dayOfWeek": ["*" ],
      "dayOfMonth": ["*" ],
      "month": [ "*” ]
  },
  "enabled": true
}
Classification
{
  "batchJobType": "monitoring",
  "deploymentId": "<deployment_id>",
  "intakeSettings": {
      "type": "jdbc",
      "dataStoreId": "<data_store_id>",
      "credentialId": "<credential_id>",
      "table": "lending_club_regression",
      "schema": "SCORING_CODE_UDF_SCHEMA",
      "catalog": "SANDBOX"
  },
  "outputSettings": {
      "type": "jdbc",
      "dataStoreId": "<data_store_id>",
      "table": "lending_club_regression_out",
      "catalog": "SANDBOX",
      "schema": "SCORING_CODE_UDF_SCHEMA",
      "statementType": "insert",
      "createTableIfNotExists": true,
      "credentialId": "<credential_id>",
      "commitInterval": 10,
      "whereColumns": [],
      "updateColumns": []
  },
  "monitoringColumns": {
  "predictionsColumns": [
              {
                "className": "True",
                "columnName": "readmitted_True_PREDICTION"
              },
              {
                "className": "False",
                "columnName": "readmitted_False_PREDICTION"
              }
          ],
      "associationIdColumn": "id",
      "actualsValueColumn": "loan_amnt"
  },
  "monitoringOutputSettings": {
      "uniqueRowIdentifierColumns": ["id"],
      "monitoredStatusColumn": "monitored"
  }
  "schedule": {
      "minute": [ 0  ],
      "hour": [ 17   ],
      "dayOfWeek": ["*" ],
      "dayOfMonth": ["*" ],
      "month": [ "*” ]
  },
  "enabled": true
}
Updated September 6, 2024