public Wrapper(IDataFeatureProvider baseProvider, IEnumerable<int> indexes)
{
Contract.Requires(baseProvider != null);
Contract.Requires(indexes != null);
this.BaseProvider = baseProvider;
this.indexes = indexes.OrderBy(i => i).Select((i, idx) => new { I = i, Idx = idx }).ToDictionary(info => info.Idx, info => info.I);
}
void IValidatorFactory.GetProviders(out IDataFeatureProvider trainingProvider, out IDataFeatureProvider validationProvider)
{
Contract.Ensures(Contract.ValueAtReturn<IDataFeatureProvider>(out trainingProvider) != null);
Contract.Ensures(Contract.ValueAtReturn<IDataFeatureProvider>(out validationProvider) != null);
Contract.Ensures(Contract.ValueAtReturn<IDataFeatureProvider>(out trainingProvider).FeatureDescriptions ==
Contract.ValueAtReturn<IDataFeatureProvider>(out validationProvider).FeatureDescriptions);
trainingProvider = null;
validationProvider = null;
}
// TODO: Configurable Cache!
public DataFeatureMatrixProvider(DataFeatureSelectionStrategy selectionStrategy, IDataFeatureProvider dataFeatureProvider, Strings featureIDs = null)
{
Contract.Requires(selectionStrategy != null);
Contract.Requires(dataFeatureProvider != null);
this.selectionStrategy = selectionStrategy;
DataFeatureProvider = dataFeatureProvider;
vectorizer = new FeatureSetVectorizer(featureIDs == null ? dataFeatureProvider.FeatureDescriptions : dataFeatureProvider.FeatureDescriptions.GetSubset(featureIDs));
MatrixWidth = vectorizer.FeatureDescriptions.Sum(d => d.FeatureValueCount);
}
public ClientValidatorFactory(IDataFeatureProvider baseProvider, int numberOfValidationItems, ValidationSamplingMethod method = ValidationSamplingMethod.FromStart)
{
Contract.Requires(baseProvider != null);
Contract.Requires(numberOfValidationItems > 0);
BaseProvider = baseProvider;
NumberOfValidationItems = numberOfValidationItems;
NumberOfTrainingItems = baseProvider.ItemCount - numberOfValidationItems;
Method = method;
if (NumberOfTrainingItems <= 0) throw new ArgumentException("Too many validation item count.", "numberOfValidationItems");
}
public void GetProviders(out IDataFeatureProvider trainingProvider, out IDataFeatureProvider validationProvider)
{
try
{
trainingProvider = BaseProvider.GetDataSubsetProvider(TrainingProviderSubsetParameters);
validationProvider = BaseProvider.GetDataSubsetProvider(ValidationProviderSubsetParameters);
}
catch (Exception ex)
{
throw new InvalidOperationException("Cannot create subset providers. See inner exception for details.", ex);
}
}
public void GetProviders(out IDataFeatureProvider trainingProvider, out IDataFeatureProvider validationProvider)
{
if (Method == ValidationSamplingMethod.FromStart)
{
validationProvider = new Wrapper(BaseProvider, Enumerable.Range(0, NumberOfValidationItems));
trainingProvider = new Wrapper(BaseProvider, Enumerable.Range(NumberOfValidationItems, NumberOfTrainingItems));
}
else if (Method == ValidationSamplingMethod.FromEnd)
{
trainingProvider = new Wrapper(BaseProvider, Enumerable.Range(0, NumberOfTrainingItems));
validationProvider = new Wrapper(BaseProvider, Enumerable.Range(NumberOfTrainingItems, NumberOfValidationItems));
}
else
{
var validationIndexes = new HashSet<int>();
do { validationIndexes.Add(RandomGenerator.Random.Next(BaseProvider.ItemCount)); } while (validationIndexes.Count != NumberOfValidationItems);
validationProvider = new Wrapper(BaseProvider, validationIndexes);
trainingProvider = new Wrapper(BaseProvider, Enumerable.Range(0, BaseProvider.ItemCount).Except(validationIndexes));
}
}