Provides standard ways of combining numeric test results from a number of partitions. These can be used in the combine each phase of k-fold cross validation.

Declaration

public static double Combine(IEnumerable<ICrossValidationTestResult> testResults, CrossValidationCombineMethod method)

Parameters

Type	Name	Description
IEnumerable<ICrossValidationTestResult>	testResults	The test results. Typically one for each test partitioning.
CrossValidationCombineMethod	method	The method to be used to combine the test results.

Returns

Type	Description
Double	The overall cross validation score, which can be used to compare models.

kFold(Int32, Int32, Func<DataPartitioning, ICrossValidationNetwork>, Func<DataPartitioning, ICrossValidationNetwork, ICrossValidationTestResult[]>, Func<Int32, ICrossValidationTestResult[], ICrossValidationScore>)

Performs k-fold cross validation.

Declaration

public static ICrossValidationScore[] kFold(int partitionCount, int testMetricCount, Func<DataPartitioning, ICrossValidationNetwork> learn, Func<DataPartitioning, ICrossValidationNetwork, ICrossValidationTestResult[]> test, Func<int, ICrossValidationTestResult[], ICrossValidationScore> combine)

Parameters

Type	Name	Description
Int32	partitionCount	The number of partitions to use.
Int32	testMetricCount	The number of test metrics being calculated.
Func<DataPartitioning, ICrossValidationNetwork>	learn	A user supplied function to learn a network based on a training partitioning of the data.
Func<DataPartitioning, ICrossValidationNetwork, ICrossValidationTestResult[]>	test	A user supplied function to test the network on a test partioning of the data.
Func<Int32, ICrossValidationTestResult[], ICrossValidationScore>	combine	A user supplied function to combine the test results over multiple partitionings.

Returns

Type	Description
ICrossValidationScore[]	A score for each test metric.

kFoldList(Int32)

Gets a list of training and test DataPartitioning instances for each partition.

Declaration

public static IList<CrossValidationOutput> kFoldList(int partitionCount)

Parameters

Type	Name	Description
Int32	partitionCount	The number of k-fold partitions.

Returns

Type	Description
IList<CrossValidationOutput>	Training and test DataPartitioning instances for each partition.

Remarks

This method is an alternative to the callback based BayesServer.Data.CrossValidation.kFold method.