datarobot-mlops

DataRobot MLOps metrics reporting library

This is the Python version of the DataRobot MLOps reporting SDK. This library enables remote reporting of MLOps metrics back to DataRobot for monitoring.

More information on DataRobot MLOps reporting library can be found here: https://docs.datarobot.com/en/docs/mlops/deployment/mlops-agent/index.html.

Installation

Install the MLOps reporting SDK with minimal dependencies. This means that only the filesystem spooler type will be available.

pip install datarobot-mlops

By installing additional dependencies, you can leverage these additional spooler types:

• rabbitmq - Installs dependencies for the rabbitmq spooler type.
• google - Installs dependencies for the pubsub spooler type.
• kafka - Installs dependencies for the kafka spooler type.
• azure - Installs dependencies that enable Azure Active Directory authentication for the kafka spooler type.
• aws - Installs dependencies for the sqs spooler type.
• api - Installs dependencies for the api spooler type.
• aggregator - Installs dependencies for large-scale monitoring

You can install these extra dependencies using pip, for example:

pip install 'datarobot-mlops[rabbitmq]'

You can also install multiple sets of dependencies at once, for example:

pip install 'datarobot-mlops[kafka,azure]'