public double Expect(Col other)
{
var n = N + other.N;
return(N / n * Variety() + other.N / n * other.Variety());
}
public void Read(string fileName)
{
int numIgnoreCols = 0;
using (var reader = new StreamReader(fileName))
{
int row = 0;
while (!reader.EndOfStream)
{
var line = reader.ReadLine().Trim();
var values = line.Split(',');
if (Cols.Count > 0 && Cols.Count() != values.Count()) //toss mismatched
{
Console.Error.WriteLine("Possible Error on line " + row);
}
if (row == 0)
{
int colNum = 1;
//we have a well-defined header
foreach (var c in values)
{
var newCol = Col.Create(c.Trim());
newCol.Index = colNum;
colNum++;
Cols.Add(newCol);
if (newCol.Ignore)
{
numIgnoreCols++;
}
}
row++;
continue;
}
int col = 0;
var currRow = new Row();
int numColValues = 0;
foreach (var c in values)
{
if (Cols[col].Ignore)
{
col++;
continue;
}
//TODO don't let both the Row and Cols hang onto the Cells. I think I would dump the Row Construct?
var cell = currRow.AddCell(c.Trim(), Cols[col].CellType, Cols[col], row);
if (!cell.IsEmpty())
{
numColValues++;
Cols[col].AddCell(cell);
}
col++;
}
row++;
Rows.Add(currRow);
int predictor = 0;
foreach (var predict in _predicts)
{
var prediction = predict.AddRow(currRow, Cols);
if (prediction != null)
{
_predictReporter[predictor].RecordPrediction(currRow.GetGoals().First().RawValue, prediction);
}
predictor++;
}
}
}
}
public abstract bool IsCentralTendencyDifferent(Col other, double epsilon);
