.. _model-surrogate-global-metrics-cross_validation:

""""""""""""""""
cross_validation
""""""""""""""""


Perform k-fold cross validation



**Topics**


surrogate_models


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

   model-surrogate-global-metrics-cross_validation-folds
   model-surrogate-global-metrics-cross_validation-percent


**Specification**

- *Alias:* None

- *Arguments:* None

- *Default:* No cross validation


**Child Keywords:**

+-------------------------+--------------------+--------------------+-----------------------------------------------+
| Required/Optional       | Description of     | Dakota Keyword     | Dakota Keyword Description                    |
|                         | Group              |                    |                                               |
+=========================+====================+====================+===============================================+
| Optional (Choose One)   | Partition Control  | `folds`__          | Number of cross validation folds              |
|                         |                    +--------------------+-----------------------------------------------+
|                         |                    | `percent`__        | Percent data per cross validation fold        |
+-------------------------+--------------------+--------------------+-----------------------------------------------+

.. __: model-surrogate-global-metrics-cross_validation-folds.html
__ model-surrogate-global-metrics-cross_validation-percent.html



**Description**


General k-fold cross validation may be performed by specifying
``cross_validation``.  The cross-validation statistics will be calculated
for all metrics.

Cross validation may further specify:

- ``folds``, the number of folds into which to divide the build   data (between 2 and number of data points) or
- ``percent``, the fraction of data (between 0 and 0.5)   to use in each fold.

These will be adjusted as needed based on the number of available
training points.  The default number of folds k = 10, or 0.1