.. _model-surrogate-global_approx-dace_method_pointer-auto_refinement-cross_validation_metric:

"""""""""""""""""""""""
cross_validation_metric
"""""""""""""""""""""""


Choice of error metric to satisfy


.. toctree::
   :hidden:
   :maxdepth: 1

   model-surrogate-global_approx-dace_method_pointer-auto_refinement-cross_validation_metric-folds


**Specification**

- *Alias:* None

- *Arguments:* STRING

- *Default:* root-mean-squared error


**Child Keywords:**

+-------------------------+--------------------+--------------------+-----------------------------------------------+
| Required/Optional       | Description of     | Dakota Keyword     | Dakota Keyword Description                    |
|                         | Group              |                    |                                               |
+=========================+====================+====================+===============================================+
| Optional                                     | `folds`__          | Number of cross validation folds              |
+----------------------------------------------+--------------------+-----------------------------------------------+

.. __: model-surrogate-global_approx-dace_method_pointer-auto_refinement-cross_validation_metric-folds.html



**Description**


The cross-validation score that the auto-refinement routine seeks
to satisfy is based on an error metric. Any of the surrogate diagnostic
:dakkw:`model-surrogate-global-metrics` may be used. These include:


- root_mean_squared

- mean_abs

- rsquared

- sum_squared

- mean_squared

- sum_abs

- max_abs

*Default Behavior*
The default metric is the the root mean squared error ('root_mean_squared').


