Sklearn Stacking ClassifierInstall

Summary

Predict labels and probabilities using a pre‑trained stacking classifier model.

Inputs

• modelData (required) — JSON string containing the serialized model trained in the 'train' operation.
• predictFeatureColumns (required) — Comma‑separated list of feature column names that correspond to columns used during training.
• pythonPath — Path to the Python executable that has scikit‑learn installed; defaults to 'python3'.

Output shape

an array of workflow items, each with a predictions field (list of predicted class labels) and a probabilities field (list of class probability vectors)

Feature columns must match those used when training the model; missing columns cause an error. Empty output occurs only on successful execution.

Examples

Example 1: Run prediction on a single record

modelData: '{"model_pickle": "...","score": 0.85,...}', predictFeatureColumns: 'age,income', pythonPath: 'python3'

Example 2: Predict on a batch of rows

modelData: '', predictFeatureColumns: 'feature1,feature2,feature3', pythonPath: '/usr/bin/python3'

Example 3: Use a custom Python interpreter

modelData: '', predictFeatureColumns: 'col_a,col_b', pythonPath: '/opt/conda/envs/ml/bin/python'