public IEnumerable <(IDataTable Training, IDataTable Validation)> Fold(int k, int?randomSeed = null, bool shuffle = true)
{
var input = Enumerable.Range(0, RowCount);
if (shuffle)
{
input = input.Shuffle(randomSeed);
}
var final = input.ToList();
var foldSize = final.Count / k;
for (var i = 0; i < k; i++)
{
var trainingRows = final.Take(i * foldSize).Concat(final.Skip((i + 1) * foldSize));
var validationRows = final.Skip(i * foldSize).Take(foldSize);
var writer1 = new DataTableWriter(Columns, null);
foreach (var row in GetRows(trainingRows))
{
writer1.Process(row);
}
var writer2 = new DataTableWriter(Columns, null);
foreach (var row in GetRows(validationRows))
{
writer2.Process(row);
}
yield return(writer1.GetDataTable(), writer2.GetDataTable());
}
}
public (IDataTable Training, IDataTable Test) Split(int?randomSeed = null, double trainPercentage = 0.8, bool shuffle = true, Stream output1 = null, Stream output2 = null)
{
var input = Enumerable.Range(0, RowCount);
if (shuffle)
{
input = input.Shuffle(randomSeed);
}
var final = input.ToList();
int trainingCount = Convert.ToInt32(RowCount * trainPercentage);
var writer1 = new DataTableWriter(Columns, output1);
foreach (var row in GetRows(final.Take(trainingCount)))
{
writer1.Process(row);
}
var writer2 = new DataTableWriter(Columns, output2);
foreach (var row in GetRows(final.Skip(trainingCount)))
{
writer2.Process(row);
}
return(writer1.GetDataTable(), writer2.GetDataTable());
}
public IDataTable Bag(int?count = null, Stream output = null, int?randomSeed = null)
{
var input = Enumerable.Range(0, RowCount).ToList().Bag(count ?? RowCount, randomSeed);
var writer = new DataTableWriter(Columns, output);
foreach (var row in GetRows(input))
{
writer.Process(row);
}
return(writer.GetDataTable());
}