.. _auto_ml_predict_info:

*ML INTERACTIVE PREDICTION*
===========================

Loads a previously trained ML model file (single .pkl, comma-separated .pkl list, or per-target CSV manifest) and interrogates it via LucyML to surface model metadata — including inputs, targets, training data schema, and configuration — without requiring new prediction data. Use this worker to inspect or pass model context downstream in a workflow before or instead of running a full prediction.

When to use
-----------

Tagged: ``info``, ``lucy``, ``lucyml``, ``ml``, ``model inspection``, ``model metadata``, ``per-target``, ``pkl``.

Inputs
------

.. list-table::
   :header-rows: 1
   :widths: 20 20 20 20 20 20

   * - Label
     - ID
     - Type
     - Default
     - Required
     - Description
   * - Saved Model File
     - mfile
     - text
     - —
     - ✓
     - Path, numeric mathmodel ID, or inline per-target CSV manifest pointing to the saved model file (.pkl) produced by a prior training worker; also accepts a comma-separated list of .pkl paths for multi-target models.

Outputs
-------

.. list-table::
   :header-rows: 1
   :widths: 20 20 20 20

   * - Label
     - ID
     - Type
     - Description
   * - Model File
     - mfile
     - text
     - Resolved filesystem path(s) to the loaded model file(s); comma-separated when multiple per-target models were processed.
   * - Lucy JSON
     - lucy_json
     - json
     - Raw LucyML JSON envelope (dataType: lucy_ml_json) returned by the model-info call, containing the full payload of all responses from the Python backend.
   * - Independents
     - inputs
     - scalar
     - Scalar list of independent (feature) variable names registered in the model.
   * - Targets
     - targets
     - scalar
     - Scalar list of target (response) variable names the model was trained to predict; also surfaced as a dataset containing the original training target values.
   * - Training Data
     - targets
     - dataset
     - Scalar list of target (response) variable names the model was trained to predict; also surfaced as a dataset containing the original training target values.
   * - Schema
     - schema
     - dataset
     - Dataset describing the model's feature schema — column names, types, and any preprocessing metadata — as extracted from the saved model file.
   * - Config Parameters
     - config
     - keyvalue
     - Key-value map of configuration parameters (hyperparameters, algorithm settings, etc.) stored inside the model at training time.

Disciplines
-----------

- ai_ml.model_selection
- ai_ml.supervised.classification
- ai_ml.supervised.regression
- ai_ml.surrogate

.. raw:: html

   <hr style="margin-top:2em">
   <p style="font-size:11px;color:#888">
   Auto-generated from <code>platform</code> schema. Worker id: <code>ml_predict_info</code>. Schema hash: <code>b4d8d70e9ba8</code>. Hand-curated docs in <code>workerexamples/</code> override this page when present.
   </p>