airflow-provider-datarobot 0.0.5

DataRobot Airflow provider.

DataRobot Provider for Apache Airflow

This package provides operators, sensors, and a hook that integrates DataRobot into Apache Airflow. Using these components you should be able to build the essential DataRobot pipeline - create a project, train models, deploy a model, score predictions against the model deployment.

Installation

Prerequisites:

• Apache Airflow
• DataRobot Python API client

Install the DataRobot provider:

pip install airflow-provider-datarobot

Connection

In the Airflow user interface, create a new DataRobot connection in Admin > Connections:

• Connection Type: DataRobot
• Connection Id: datarobot_default (default)
• API Key: your-datarobot-api-key
• DataRobot endpoint URL: https://app.datarobot.com/api/v2 (default)

Create the API Key in the DataRobot Developer Tools page, API Keys section (see DataRobot Docs for more details).

By default, all components use datarobot_default connection ID.

Config JSON for dag run

Operators and sensors use parameters from the config which must be submitted when triggering the dag. Example config JSON with required parameters:

{
    "training_data": "s3-presigned-url-or-local-path-to-training-data",
    "project_name": "Project created from Airflow",
    "autopilot_settings": {
        "target": "readmitted"
    },
    "deployment_label": "Deployment created from Airflow",
    "score_settings": {
        "intake_settings": {
            "type": "s3",
            "url": "s3://path/to/scoring-data/Diabetes10k.csv",
            "credential_id": "62160b511fb29da8dd5f2c81"
        },
        "output_settings": {
            "type": "s3",
            "url": "s3://path/to/results-dir/Diabetes10k_predictions.csv",
            "credential_id": "62160b511fb29da8dd5f2c81"
        }
    }
}

These config values can be accessed in the execute() method of any operator the dag in the context["params"] variable, e.g. getting a training data you would use this in the operator:

def execute(self, context: Dict[str, Any]) -> str:
    ...
    training_data = context["params"]["training_data"]
    ...

Modules

Operators

• UploadDatasetOperator

  Uploading local file to DataRobot AI Catalog and return Dataset ID.

  Required config params:

    dataset_file_path: str - local path to training dataset
  

  Returns a dataset ID.

• CreateProjectOperator

  Creates a DataRobot project and returns its ID.

  Several options of source dataset supported:

  • Creating project directly from local file or pre-signed S3 URL. Required config params:

     training_data: str - pre-signed S3 URL or local path to training dataset
     project_name: str - project name
    

    In case of an S3 input, the training_data value must be a pre-signed AWS S3 URL.

    For more project settings see the DataRobot docs.

    Returns a project ID.

  • Creating project from existing dataset in AI Catalog, using dataset_id from config file. Required config params:

     training_dataset_id: str - dataset_id corresponding to existing dataset in DataRobot AICatalog
     project_name: str - project name
    

    For more project settings see the DataRobot docs.

    Returns a project ID.

  • Creating project from an existing dataset in DataRobot AI Catalog, using dataset_id coming from previous operator. In this case your previous operator must return valid dataset_id (for example UploadDatasetOperator) and you should use this output value as a 'dataset_id' argument in CreateProjectOperator object creation step. Required config params:

     project_name: str - project name
    

    For more project settings see the DataRobot docs.

    Returns a project ID.

• TrainModelsOperator

  Triggers DataRobot Autopilot to train models.

  Parameters:

    project_id: str - DataRobot project ID
  

  Required config params:

    "autopilot_settings": {
        "target": "readmitted"
    } 
  

  target is a required parameter with the column name which defines the modeling target.

  For more autopilot settings see the DataRobot docs.

  Returns None.

• DeployModelOperator

  Deploys a specified model.

  Parameters:

    model_id: str - DataRobot model ID
  

  Required config params:

    deployment_label - deployment label/name
  

  For more deployment settings see the DataRobot docs.

  Returns a deployment ID.

• DeployRecommendedModelOperator

  Deploys a recommended model.

  Parameters:

    project_id: str - DataRobot project ID
  

  Required config params:

    deployment_label: str - deployment label
  

  For more deployment settings see the DataRobot docs.

  Returns a deployment ID.

• ScorePredictionsOperator

  Scores batch predictions against the deployment.

  Prerequisites:

  • S3 credentials added to DataRobot via Python API client. You need the creds.credential_id for the credential_id parameter in the config.
  • OR a Dataset ID in the AI Catalog

  Parameters:

    deployment_id: str - DataRobot project ID
  

  Required config params:

    "score_settings": {
        "intake_settings": {
            "type": "s3",
            "url": "s3://my-bucket/Diabetes10k.csv",
            "credential_id": "62160b511fb29da8dd5f2c81"
        },
        "output_settings": {
            "type": "s3",
            "url": "s3://my-bucket/Diabetes10k_predictions.csv",
            "credential_id": "62160b511fb29da8dd5f2c81"
        }
    }
  

  Config params for scoring a Dataset in the AI Catalog:

    "score_settings": {
        "intake_settings": {
            "type": "dataset",
            "dataset_id": "<datasetId>",
        },
        "output_settings": {
            ...
        }
    }
  

  For more batch prediction settings see the DataRobot docs.

  Returns a batch prediction job ID.

• GetTargetDriftOperator

  Gets the target drift from a deployment.

  Parameters:

    deployment_id: str - DataRobot deployment ID
  

  No config params are required. Optional params may be passed in the config as follows:

    "target_drift": {
        ...
    }
  

  Returns a dict with the target drift data.

• GetFeatureDriftOperator

  Gets the feature drift from a deployment.

  Parameters:

    deployment_id: str - DataRobot deployment ID
  

  No config params are required. Optional params may be passed in the config as follows:

    "feature_drift": {
        ...
    }
  

  Returns a dict with the feature drift data.

Sensors

• AutopilotCompleteSensor

  Checks whether the Autopilot has completed.

  Parameters:

    project_id: str - DataRobot project ID
  

• ScoringCompleteSensor

  Checks whether batch scoring has completed.

  Parameters:

    job_id: str - Batch prediction job ID
  

Hooks

• DataRobotHook

  A hook for initializing DataRobot Public API client.

Pipeline

The modules described above allows to construct a standard DataRobot pipeline in an Airflow dag:

create_project_op >> train_models_op >> autopilot_complete_sensor >> deploy_model_op >> score_predictions_op >> scoring_complete_sensor

Examples

See the examples directory for the example DAGs.

Issues

Please submit issues and pull requests in our official repo: https://github.com/datarobot/airflow-provider-datarobot

We are happy to hear from you. Please email any feedback to the authors at [e-mail].

Copyright Notice

Copyright 2023 DataRobot, Inc. and its affiliates.

All rights reserved.

This is proprietary source code of DataRobot, Inc. and its affiliates.

Released under the terms of DataRobot Tool and Utility Agreement.