Sklearn Calibrated Classifier icon

Sklearn Calibrated Classifier

Probability calibration for classifiers using scikit-learn

Actions2

Train

AI-generated

Summary

Train a calibrated classifier using scikit-learn on specified feature and target columns.

Inputs

  • featureColumns (required) — Comma-separated list of feature column names
  • targetColumn (required) — Name of the target column
  • baseEstimator — Base classifier to calibrate (e.g., SVC, RandomForestClassifier)
  • method — Calibration method (sigmoid or isotonic)
  • cv — Number of cross-validation folds (default 5)
  • pythonPath — Path to Python executable with scikit-learn installed (default python3)

Output shape

a JSON object containing model_pickle (base64 string), base_estimator (string), method (string), classes (array), feature_columns (array) and training_samples (number)

outputs a single record per workflow item; training_samples shows how many training rows were used

Examples

Example 1: Train an SVC with sigmoid calibration on columns feature1,feature2,feature3 targeting 'target'

featureColumns='feature1,feature2,feature3', targetColumn='target', baseEstimator='SVC', method='sigmoid', cv=5, pythonPath='python3'

Example 2: Predict using the trained model on new rows

modelData='', predictFeatureColumns='feature1,feature2,feature3'

Discussion