Skip to content

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

Prediction intake options

You can configure a prediction source using the Predictions > Job Definitions tab or the Batch Prediction API. This topic describes both the UI and API intake options.

Note

For a complete list of supported intake options, see the data sources supported for batch predictions.

Intake option Description
Local file streaming Stream input data through a URL endpoint for immediate processing when the job moves to a running state.
HTTP scoring Stream input data from an absolute URL for scoring. This option can read data from pre-signed URLs for Amazon S3, Azure, and Google Cloud Platform.
Database connections
JDBC scoring Read prediction data from a JDBC-compatible database with data source details supplied through a job definition or the Batch Prediction API.
SAP Datasphere scoring Read prediction data from a SAP Datasphere database with data source details supplied through a job definition or the Batch Prediction API.
Cloud storage connections
Azure Blob Storage scoring Read input data from Azure Blob Storage with DataRobot credentials consisting of an Azure Connection String.
Google Cloud Storage scoring (GCP) Read input data from Google Cloud Storage with DataRobot credentials consisting of a JSON-formatted account key.
Amazon S3 scoring Read input data from public or private S3 buckets with DataRobot credentials consisting of an access key (ID and key) and a session token (Optional) This is the preferred intake option for larger files.
Data warehouse connections
BigQuery scoring Score data using BigQuery with data source details supplied through a job definition or the Batch Prediction API.
Snowflake scoring Score data using Snowflake with data source details supplied through a job definition or the Batch Prediction API.
Synapse scoring Score data using Synapse with data source details supplied through a job definition or the Batch Prediction API.
Other connections
AI Catalog / Data Registry dataset scoring Read input data from a dataset snapshot in the DataRobot AI Catalog / Data Registry.
Wrangler Recipe scoring Read input data from a wrangler recipe created in the DataRobot NextGen Workbench from a Snowflake data connection.

If you are using a custom CSV format, any intake option dealing with CSV will adhere to that format.

Local file streaming

Local file intake does not have any special options. This intake option requires you to upload the job's scoring data using a PUT request to the URL specified in the csvUpload link in the job data. This starts the job (or queues it for processing if the prediction instance is already occupied).

If there is no other queued job for the selected prediction instance, scoring will start while you are still uploading.

Refer to this sample use case.

Note

If you forget to send scoring data, the job remains in the INITIALIZING state.

Multipart upload

Because the local file intake process requires that you upload scoring data for a job using a PUT request to the URL specified in the csvUpload parameter, by default, a single PUT request starts the job (or queues it for processing if the prediction instance is occupied). Multipart upload for batch predictions allows you to override the default behavior to upload scoring data through multiple files. This upload process requires multiple PUT requests followed by a single POST request (finalizeMultipart) to finalize the multipart upload manually. This feature can be helpful when you want to upload large datasets over a slow connection or if you experience frequent network instability.

Note

For more information on the batch prediction API and local file intake, see Batch Prediction API and Prediction intake options.

Multipart upload endpoints

This feature adds the following multipart upload endpoints to the batch prediction API:

Endpoint Description
PUT /api/v2/batchPredictions/:id/csvUpload/part/0/ Upload scoring data in multiple parts to the URL specified by csvUpload. Increment 0 by 1 in sequential order for each part of the upload.
POST /api/v2/batchPredictions/:id/csvUpload/finalizeMultipart/ Finalize the multipart upload process. Make sure each part of the upload has finished before finalizing.

Local file intake settings

The intake settings for the local file adapter added two new properties to support multipart upload for the batch prediction API:

Property Type Default Description
intakeSettings.multipart boolean false
  • true: Requires you to submit multiple files via a PUT request and finalize the process manually via a POST request (finalizeMultipart).
  • false: Finalizes intake after one file is submitted via a PUT request.
intakeSettings.async boolean true
  • true: Starts the scoring job when the initial PUT request for file intake is made.
  • false: Postpones the scoring job until the PUT request resolves or the POST request for finalizeMultipart resolves.
Multipart intake setting

To enable the new multipart upload workflow, configure the intakeSettings for the localFile adapter as shown in the following sample request:

{
    "intakeSettings": {
        "type": "localFile",
        "multipart": true
    }
}
These properties alter the local file upload workflow, requiring you to:

  • Upload any number of sequentially numbered files.

  • Finalize the upload to indicate that all required files uploaded successfully.

Async intake setting

To enable the new multipart upload workflow with async enabled, configure the intakeSettings for the localFile adapter as shown in the following sample request:

Note

You can also use the async intake setting independently of the multipart setting.

{
    "intakeSettings": {
        "type": "localFile",
        "multipart": true,
        "async": false
    }
}

A defining feature of batch predictions is that the scoring job starts on the initial file upload, and only one batch prediction job at a time can run for any given prediction instance. This functionality may cause issues when uploading large datasets over a slow connection. In these cases, the client's upload speed could create a bottleneck and block the processing of other jobs. To avoid this potential bottleneck, you can set async to false, as shown in the example above. This configuration postpones submitting the batch prediction job to the queue.

When "async": false, the point at which a job enters the batch prediction queue depends on the multipart setting:

  • If "multipart": true, the job is submitted to the queue after the POST request for finalizeMultipart resolves.

  • If "multipart": false, the job is submitted to the queue after the initial file intake PUT request resolves.

Example multipart upload requests

The batch prediction API requests required to upload a 3 part multipart batch prediction job would be:

PUT /api/v2/batchPredictions/:id/csvUpload/part/0/

PUT /api/v2/batchPredictions/:id/csvUpload/part/1/

PUT /api/v2/batchPredictions/:id/csvUpload/part/2/

POST /api/v2/batchPredictions/:id/csvUpload/finalizeMultipart/

Each uploaded part is a complete CSV file with a header.

Abort a multipart upload

If you start a multipart upload that you don't want to finalize, you can use a DELETE request to the existing batchPredictions abort route:

DELETE /api/v2/batchPredictions/:id/

HTTP scoring

In addition to the cloud storage adapters, you can also point batch predictions to a regular URL so DataRobot can stream the data for scoring:

Parameter Example Description
type http Use HTTP for intake.
url https://example.com/datasets/scoring.csv An absolute URL for the file to be scored.

The URL can optionally contain a username and password, such as https://username:password@example.com/datasets/scoring.csv.

The http adapter can be used for ingesting data from pre-signed URLs from either S3, Azure, or GCP.

JDBC scoring

DataRobot supports reading from any JDBC-compatible database for Batch Predictions. To use JDBC with the Batch Prediction API, specify jdbc as the intake type. Since no file is needed for a PUT request, scoring will start immediately, transitioning the job to RUNNING if preliminary validation succeeds. To support this, the Batch Prediction API integrates with external data sources using credentials securely stored in data credentials.

Supply data source details using the Predictions > Job Definitions tab or the Batch Prediction API (intakeSettings) as described in the table below.

UI field Parameter Example Description
Source type type jdbc Use a JDBC data store for intake.
Data connection options
+ Select connection dataStoreId 5e4bc5b35e6e763beb9db14a The ID of an external data source. In the UI, select a data connection or click add a new data connection. Complete account and authorization fields.
Enter credentials credentialId 5e4bc5555e6e763beb9db147 The ID of a stored credential. Refer to storing credentials securely.
Schemas schema public (Optional) The name of the schema containing the table to be scored.
Tables table scoring_data (Optional) The name of the database table containing data to be scored.
SQL query query SELECT feature1, feature2, feature3 AS readmitted FROM diabetes (Optional) A custom query to run against the database.
Deprecated option
Fetch size fetchSize (deprecated) 1000 Deprecated: fetchSize is now inferred dynamically for optimal throughput and is no longer needed. (Optional) To balance throughput and memory usage, sets a custom fetchSize (number of rows read at a time). Must be in range [1, 100000]; default 1000.

Note

You must specify either table and schema or query.

Refer to the example section for a complete API example.

Data warehouse connections

Using JDBC to transfer data can be costly in terms of IOPS (input/output operations per second) and expense for data warehouses. The data warehouse adapters reduce the load on database engines during prediction scoring by using cloud storage and bulk insert to create a hybrid JDBC-cloud storage solution. For more information, see the BigQuery, Snowflake, and Synapse data warehouse adapter sections.

Allowed source IP addresses

Any connection initiated from DataRobot originates from one of the following IP addresses:

Host: https://app.datarobot.com Host: https://app.eu.datarobot.com Host: https://app.jp.datarobot.com
100.26.66.209 18.200.151.211 52.199.145.51
54.204.171.181 18.200.151.56 52.198.240.166
54.145.89.18 18.200.151.43 52.197.6.249
54.147.212.247 54.78.199.18
18.235.157.68 54.78.189.139
3.211.11.187 54.78.199.173
52.1.228.155 18.200.127.104
3.224.51.250 34.247.41.18
44.208.234.185 99.80.243.135
3.214.131.132 63.34.68.62
3.89.169.252 34.246.241.45
3.220.7.239 52.48.20.136
52.44.188.255
3.217.246.191

Note

These IP addresses are reserved for DataRobot use only.

SAP Datasphere scoring

Premium

Support for SAP Datasphere is off by default. Contact your DataRobot representative or administrator for information on enabling the feature.

Feature flag(s): Enable SAP Datasphere Connector, Enable SAP Datasphere Batch Predictions Integration

To use SAP Datasphere for scoring, supply data source details using the Predictions > Job Definitions tab or the Batch Prediction API (intakeSettings) as described in the table below.

UI field Parameter Example Description
Source type type datasphere Use a SAP Datasphere database for intake.
Data connection options
+ Select connection dataStoreId 5e4bc5b35e6e763beb9db14a The ID of an external data source. In the UI, select a data connection or click add a new data connection. Refer to the SAP Datasphere connection documentation.
Enter credentials credentialId 5e4bc5555e6e763beb9db147 The ID of a stored credential for Datasphere. Refer to storing credentials securely.
catalog / The name of the database catalog containing the table to be scored.
Schemas schema public The name of the database schema containing the table to be scored.
Tables table scoring_data The name of the database table containing data to be scored.

Azure Blob Storage scoring

A scoring option for large files is Azure. To score from Azure Blob Storage, you must configure credentials with DataRobot using an Azure Connection String.

UI field Parameter Example Description
Source type type azure Use Azure Blob Storage for intake.
URL url https://myaccount.blob.core.windows.net/datasets/scoring.csv An absolute URL for the file to be scored.
Format format csv (Optional) Select CSV (csv) or Parquet (parquet).
Default value: CSV
+ Add credentials credentialId 5e4bc5555e6e763beb488dba In the UI, enable the + Add credentials field by selecting This URL requires credentials. Required if explicit access to credentials for this URL are required (optional otherwise). Refer to storing credentials securely.

Azure credentials are encrypted and are only decrypted when used to set up the client for communication with Azure during scoring.

Google Cloud Storage scoring

DataRobot supports the Google Cloud Storage adapter. To score from Google Cloud Storage, you must set up a credential with DataRobot consisting of a JSON-formatted account key.

UI field Parameter Example Description
Source type type gcp Use Google Cloud Storage for intake.
URL url gcs://bucket-name/datasets/scoring.csv An absolute URL for the file to be scored.
Format format csv (Optional) Select CSV (csv) or Parquet (parquet).
Default value: CSV
+ Add credentials credentialId 5e4bc5555e6e763beb488dba In the UI, enable the + Add credentials field by selecting This URL requires credentials. Required if explicit access credentials for this URL are required, otherwise optional. Refer to storing credentials securely.

GCP credentials are encrypted and are only decrypted when used to set up the client for communication with GCP during scoring.

Amazon S3 scoring

For larger files, S3 is the preferred method for intake. DataRobot can ingest files from both public and private buckets. To score from Amazon S3, you must set up a credential with DataRobot consisting of an access key (ID and key) and, optionally, a session token.

UI field Parameter Example Description
Source type type s3 DataRobot recommends S3 for intake.
URL url s3://bucket-name/datasets/scoring.csv An absolute URL for the file to be scored.
Format format csv (Optional) Select CSV (csv) or Parquet (parquet).
Default value: CSV
+ Add credentials credentialId 5e4bc5555e6e763beb488dba In the UI, enable the + Add credentials field by selecting This URL requires credentials. Required if explicit access credentials for this URL are required. Refer to storing credentials securely.

AWS credentials are encrypted and only decrypted when used to set up the client for communication with AWS during scoring.

Note

If running a Private AI Cloud within AWS, it is possible to provide implicit credentials for your application instances using an IAM Instance Profile to access your S3 buckets without supplying explicit credentials in the job data. For more information, see the AWS documentation.

BigQuery scoring

To use BigQuery for scoring, supply data source details using the Predictions > Job Definitions tab or the Batch Prediction API (intakeSettings) as described in the table below.

UI field Parameter Example Description
Source type type bigquery Use the BigQuery API to unload data to Google Cloud Storage and use it as intake.
Dataset dataset my_dataset The BigQuery dataset to use.
Table table my_table The BigQuery table or view from the dataset used as intake.
Bucket bucket my-bucket-in-gcs Bucket where data should be exported.
+ Add credentials credentialId 5e4bc5555e6e763beb488dba Required if explicit access credentials for this bucket are required (otherwise optional).

In the UI, enable the + Add credentials field by selecting This connection requires credentials. Refer to storing credentials securely.

Refer to the example section for a complete API example.

Snowflake scoring

To use Snowflake as for scoring, supply data source details using the Predictions > Job Definitions tab or the Batch Prediction API (intakeSettings) as described in the table below.

UI field Parameter Example Description
Source type type snowflake Adapter type.
Data connection options
+ Select connection dataStoreId 5e4bc5b35e6e763beb9db14a ID of Snowflake data source. In the UI, select a Snowflake data connection or click add a new data connection. Complete account and authorization fields.
Enter credentials credentialId 5e4bc5555e6e763beb9db147 The ID of a stored credential for Snowflake.
Tables table SCORING_DATA (Optional) Name of the Snowflake table containing data to be scored.
Schemas schema PUBLIC (Optional) Name of the schema containing the table to be scored.
SQL query query SELECT feature1, feature2, feature3 FROM diabetes (Optional) Custom query to run against the database.
Cloud storage type cloudStorageType s3 Type of cloud storage backend used in Snowflake external stage. Can be one of 3 cloud storage providers: s3/azure/gcp. Default is s3
External stage externalStage my_s3_stage Snowflake external stage. In the UI, toggle on Use external stage to enable the External stage field.
+ Add credentials cloudStorageCredentialId 6e4bc5541e6e763beb9db15c ID of stored credentials for a storage backend (S3/Azure/GCS) used in Snowflake stage. In the UI, enable the + Add credentials field by selecting This URL requires credentials.

Refer to the example section for a complete API example.

Synapse scoring

To use Synapse for scoring, supply data source details using the Predictions > Job Definitions tab or the Batch Prediction API (intakeSettings) as described in the table below.

UI field Parameter Example Description
Source type type synapse Adapter type.
Data connection options
+ Select connection dataStoreId 5e4bc5b35e6e763beb9db14a ID of Synapse data source. In the UI, select a Synapse data connection or click add a new data connection. Complete account and authorization fields.
External data source externalDatasource my_data_source Name of the Synapse external data source.
Tables table SCORING_DATA (Optional) Name of the Synapse table containing data to be scored.
Schemas schema dbo (Optional) Name of the schema containing the table to be scored.
SQL query query SELECT feature1, feature2, feature3 FROM diabetes (Optional) Custom query to run against the database.
Enter credentials credentialId 5e4bc5555e6e763beb9db147 The ID of a stored credential for Synapse. Credentials are required if explicit access credentials for this URL are required, otherwise optional. Refer to storing credentials securely.
+ Add credentials cloudStorageCredentialId 6e4bc5541e6e763beb9db15c ID of a stored credential for Azure Blob storage. In the UI, enable the + Add credentials field by selecting This external data source requires credentials.

Refer to the example section for a complete API example.

Note

Synapse supports fewer collations than the default Microsoft SQL Server. For more information, reference the Synapse documentation.

AI Catalog / Data Registry dataset scoring

To read input data from an AI Catalog / Data Registry dataset, the following options are available:

UI field Parameter Example Description
Source type type dataset In the UI, select AI Catalog (or Data Registry in NextGen).
+ Select source from AI Catalog datasetId 5e4bc5b35e6e763beb9db14a The AI Catalog dataset ID.

In the UI, search for the dataset, select the dataset, then click Use the dataset (or Confirm in NextGen).
+ Select version datasetVersionId 5e4bc5555e6e763beb488dba The AI Catalog dataset version ID (Optional)

In the UI, enable the + Select version field by selecting the Use specific version check box. Search for and select the version. If datasetVersionId is not specified, it defaults to the latest version for the specified dataset.

Note

For the specified AI Catalog dataset, the version to be scored must have been successfully ingested, and it must be a snapshot.

Wrangler recipe dataset scoring

The following options are available to read input data from a wrangler recipe created in the DataRobot NextGen Workbench from a Snowflake data connection:

Wrangler data connection

Wrangler recipes for batch prediction jobs support data wrangled from a Snowflake data connection or the AI Catalog/Data Registry.

UI field Parameter Example Description
Source type type recipe In the UI, select Wrangler Recipe.
+ Select wrangler recipe recipeId 65fb040a42c170ee46230133 The Wrangler Recipe dataset ID.

In the NextGen prediction jobs UI, search for the wrangled dataset, select the dataset, then click Confirm.

Updated October 18, 2024