static string SelectParameter(InstanceSet instances)
{
Console.WriteLine();
Console.WriteLine($"Selecting parameter for\r\n{instances}");
Console.WriteLine($"Gini(E) = {Math.Round(Gini(instances), 2)}");
var parameters = instances.First().Parameters.Values.OrderBy(p => p.Name);
string result = null;
double max = Double.MinValue;
foreach (var parameter in parameters)
{
var gain = GiniGain(instances, parameter.Name);
Console.WriteLine($"GiniGain({parameter.Name}) = {Math.Round(gain, 2)} ");
if (gain > max)
{
max = gain;
result = parameter.Name;
}
}
Console.WriteLine($"Selected {result}");
return(result);
}
static double GiniGain(InstanceSet instances, string parameterName)
{
Console.WriteLine($" Gini(E, {parameterName}) = {Math.Round(Gini(instances, parameterName), 2)}");
Console.WriteLine(" " + String.Join("; ", Enums.Items <Value>().Select(v => $"P({parameterName}={v}) = " + P(parameterName, v, instances) + $" Gini(E{parameterName}={v}) = " +
Math.Round(Gini(instances.Where(inst => inst.Parameters[parameterName].Value == v).ToInstanceSet()), 2))));
return(Gini(instances) - Gini(instances, parameterName));
}
public Node(InstanceSet trainingInstances, Parameter?parameter = null)
{
TrainingInstances = trainingInstances;
Parameter = parameter;
}
static double Gini(InstanceSet instances, string parameterName)
{
return(Enums.Items <Value>().Select(v => P(parameterName, v, instances) * Gini(instances.Where(inst => inst.Parameters[parameterName].Value == v).ToInstanceSet())).Sum());
}
static double Gini(InstanceSet instances)
{
return(1 - Enums.Items <Class>().Select(cls => Math.Pow(P(cls, instances), 2)).Sum());
}
static double P(string parameterName, Value value, InstanceSet instances)
{
return((double)instances.Count(inst => inst.Parameters[parameterName].Value == value) / instances.Count);
}
static double P(Class cls, InstanceSet instances)
{
return(!instances.Any() ? 0 : (double)instances.Count(inst => inst.Class == cls) / instances.Count);
}