.. _auto_dataset_compute_prediction_evaluation:

*DATASET COMPUTE PREDICTION EVALUATION*
=======================================

Computes regression prediction-quality metrics (MAE, MSE, RMSE, MAPE, R²) for one or more target columns in a predictions dataset. For each target column it automatically locates the corresponding `_pred` companion column, reconstructs actuals from `_diff` offsets when needed, and returns a tabular summary of the requested metrics. Use this worker to evaluate model performance after a batch-scoring step in an ML workflow.

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

Tagged: ``MAE``, ``MAPE``, ``MSE``, ``R2``, ``RMSE``, ``batch_evaluation``, ``model_scoring``, ``prediction_evaluation``.

Inputs
------

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

   * - Label
     - ID
     - Type
     - Default
     - Required
     - Description
   * - Predictions Dataset
     - predictions_dataset
     - dataset
     - —
     - ✓
     - Tabular dataset (d3VIEW dataset type) containing both the predicted values (columns named `<target>_pred`) and, optionally, actual values or difference columns (`<target>_diff`); must have at least one row.
   * - Actual Value Columns
     - actual_value_columns
     - text
     - —
     - ✓
     - One or more column names from `predictions_dataset` that identify the target variables to evaluate; accepts bare target names as well as `<target>_pred` or `<target>_diff` suffixed variants — suffixes are stripped automatically.
   * - Metrics Requested
     - metrics_requested
     - select
     - —
     - ✓
     - One or more regression metrics to compute: MAE (Mean Absolute Error), MSE (Mean Squared Error), RMSE (Root Mean Squared Error), MAPE (Mean Absolute Percentage Error, in %; rows where actual = 0 are skipped), or R2 (coefficient of determination); select all that are needed for the evaluation report.

Outputs
-------

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

   * - Label
     - ID
     - Type
     - Description
   * - Evaluation Results
     - evaluation_results
     - dataset
     - Dataset with one row per target column and one column per requested metric, containing the scalar metric values (dimensionless or in the units of the target variable for MAE/RMSE/MSE); null is returned for any metric that cannot be computed (e.g., all-zero actuals for MAPE, or mismatched array lengths).

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

- ai_ml.model_selection
- ai_ml.supervised.regression
- data.dataset.transform
- data.statistics

Runnable example
----------------

A runnable example is registered for this worker. Open the example workflow on the d3VIEW canvas: `/api/workflow/example?id=dataset_compute_prediction_evaluation <https://www.d3view.com/api/workflow/example?id=dataset_compute_prediction_evaluation>`_

.. raw:: html

   <hr style="margin-top:2em">
   <p style="font-size:11px;color:#888">
   Auto-generated from <code>platform</code> schema. Worker id: <code>dataset_compute_prediction_evaluation</code>. Schema hash: <code>1ce7fc30d2e8</code>. Hand-curated docs in <code>workerexamples/</code> override this page when present.
   </p>