private void DoPrune()
{
var isContinuous = (Labels.ValueCount(0) < 2);
for (var row = 0; row < Features.Rows(); row++)
{
var nearest = FindKnn(Features.Row(row), new int[1] {
row
});
var with = 0;
double withSSE = 0;
var pred = new double[1];
for (var n = 0; n < nearest.Length; n++)
{
pred[0] = 0;
var targ = Labels.Row(nearest[n]);
GetOutput(FindKnn(Features.Row(nearest[n]), new int[1] {
nearest[n]
}), pred);
if (isContinuous)
{
var delta = targ[0] - pred[0];
withSSE += (delta * delta);
}
else if (pred[0] == targ[0])
{
with++;
}
}
var without = 0;
double withoutSSE = 0;
for (var n = 0; n < nearest.Length; n++)
{
pred[0] = 0;
var targ = Labels.Row(nearest[n]);
GetOutput(FindKnn(Features.Row(nearest[n]), new int[2] {
row, nearest[n]
}), pred);
if (isContinuous)
{
var delta = targ[0] - pred[0];
withoutSSE += (delta * delta);
}
else if (pred[0] == targ[0])
{
without++;
}
}
var remove = false;
if (isContinuous)
{
withSSE = Math.Sqrt(withSSE / K);
withoutSSE = Math.Sqrt(withoutSSE / K);
if (Math.Abs(withSSE - withoutSSE) <= (0.1 * withSSE))
{
remove = true;
}
}
else if (without - with >= 0)
{
remove = true;
}
if (remove)
{
// remove the row
Features.Delete(row);
Labels.Delete(row);
row--;
}
}
Console.WriteLine();
Console.WriteLine("After pruning:");
Console.WriteLine("Number of instances: " + Features.Rows());
}