AutoML Reuse Model

What it does

This node loads a model persisted by AutoML Store Model and appends its predictions to new data. It reuses the stored preprocessing, so the input only needs the same feature columns — no retraining. Output is the input table plus a prediction column, and optionally one probability column per class. Runs on a kernel.

Inputs

One table containing every feature column the stored model was trained on; extra columns are passed through untouched. The target column is not needed. Missing feature columns raise an error.

Settings

Model — pick a stored model from the artifact picker, or type its name to override the picker (useful when the model is published by a flow that hasn’t run yet in this session). The picker lists the global artifact store — the same one AutoML Store Model publishes to. Prediction column names the appended column; for classification the original class labels are restored. Turn on class probabilities to also append <prediction>_proba_<class> columns — classification only, and only for models that expose probabilities.