Arize AI
Search…
Python SDK
Arize AI for Model Monitoring, Troubleshooting, and Explainability
Use the arize Python library to monitor machine learning predictions with a few lines of code. You can instrument it on a real-time steaming prediction pipeline as well as in a bulk notebook environment.

Installing the package

First step is installing the Arize SDK.
1
pip install arize
Copied!
Arize SDK requires python >= 3.6

Instrumenting your model

  • arize.init - initialize the client used to publish predictions
  • arize.pandas - Send over a pandas dataframe to Arize

What gets published

All features, predicted labels, and actual labels passed into the client get published to the Arize platform. Actual labels are mapped back to predictions via the prediction id, so it's important to keep record of prediction ids sent along at inference time in order to successfully match the actual label for precision and other calculated metrics.
Questions? Email us at [email protected] or Slack us in the #arize-support channel