Sklearn Calibrated ClassifierInstall

Summary

Predict calibrated probabilities for incoming data using a previously trained model.

Inputs

• modelData (required) — JSON string containing the trained model (base64-encoded pickle).
• predictFeatureColumns (required) — Comma-separated list of feature column names to use for prediction.
• pythonPath — Path to Python executable with scikit-learn installed (default: python3).

Output shape

a list of records, each extending the original input with 'prediction' and 'calibrated_probabilities' fields

Both fields are arrays; the operation appends these predictions to each workflow item.

Examples

Example 1: Add predictions to a dataset row

modelData: base64-encoded model JSON; predictFeatureColumns: 'feature1,feature2'; pythonPath: python3

Example 2: Predict on a single input item

modelData set to previously saved model, predictFeatureColumns set to relevant column names, pythonPath unchanged.