Actuals

This section explains how to submit actuals for predictions of a Deployment, which are used to calculate performance metrics.

Submitting actuals

By submitting actions you add labels to predictions made in the past. This can be used as a reference, or as training data for a next iteration of the model.

In order to submit actuals you will be using prediction log IDs. These can be retrieved using the prediction log methods.

from deeploy import CreateActuals

actuals_input: CreateActuals = {
    "prediction_ids": [
      "a6d8c781-2526-4e03-9b43-4c1a62d064db",
      "b6d8c781-2526-4e03-9b43-4c1a62d064db",
    ],
    "actual_values": [
      { "predictions": [True] },
      { "predictions": [False] },
    ],
}

actuals_response = client.upload_actuals(deployment_id, actuals_input)