public Node <T> BuildTree(List <T> rows)
{
OptimalSolution <T> solution = this.FindBestPartition(rows);
if (solution.GetGain() == 0)
{
return(new Node <T>(this.CountLabelDistribution(rows)));
}
List <T>[] partition = this.Partition(rows, solution.GetQuery());
Node <T> trueBranch = BuildTree(partition[0]);
Node <T> falseBranch = BuildTree(partition[1]);
return(new Node <T>(solution.GetQuery(), trueBranch, falseBranch));
}
public OptimalSolution <T> FindBestPartition(List <T> rows)
{
OptimalSolution <T> solution = new OptimalSolution <T>();
double bestGain = 0;
Query <T> bestQuery = null;
double currentImpurity = this.Gini(rows);
Console.WriteLine("Rows: " + rows.Count + "\nGini: " + currentImpurity);
int attributesN = rows.ElementAt(0).getAttributes().Length - 1;
for (int i = 0; i < attributesN; i++)
{
List <String> attributesValue = new List <string>();
foreach (T row in rows)
{
String val = row.getAttributes()[i];
Query <T> query = new Query <T>(i, val);
List <T>[] partition = this.Partition(rows, query);
if (partition[0].Count > 0 && partition[1].Count > 0)
{
double gain = InformationGain(partition[0], partition[1], currentImpurity);
if (gain >= bestGain)
{
bestGain = gain;
bestQuery = query;
solution.SetGain(gain);
solution.SetQuery(query);
}
}
}
}
return(solution);
}