public static void Main(string[] args)
{
Process p = Process.GetCurrentProcess();
ShowWindow(p.MainWindowHandle, 3); //SW_MAXIMIZE = 3
Console.WriteLine("reading train data...");
StarterClass s = new StarterClass();
DataTable dt = s.readCSVFile("zoo.data");
List <string> autompg_data = new List <string>();
Console.WriteLine("Reading and cleaning train data\n...");
DataHolder dh = new DataHolder(dt, null);
Console.WriteLine("finished reading and cleaning data\nBuilding tree...");
SimpleNode node = s.buildtree(null, dh, null, null);
Console.WriteLine("built tree\nreading test data...");
// validation begins
DataTable autompg_test = s.readCSVFile("zoo.test");
//object classifiedNode = null;
Console.WriteLine("finished reading and cleaning test\nstarting classification...");
Dictionary <int, string> testingNodesList = new Dictionary <int, string>();
Dictionary <int, string> classifiedNodesList = new Dictionary <int, string>();
int i = 0;
foreach (DataRow datarow in autompg_test.Rows)
{
testingNodesList[i] = (string)datarow[datarow.ItemArray.Length - 1];
var classifiedNode = s.classify(datarow, node);
classifiedNodesList[i] = classifiedNode is SimpleNode ? (((SimpleNode)classifiedNode).Name) : (string)classifiedNode;
Console.WriteLine(string.Join(",", datarow.ItemArray) + " " + classifiedNodesList[i]);
i++;
}
Console.Write("\nfinished classifying\ncalculating accuracy...");
int predictedCorrectValues = 0;
for (int m = 0; m < testingNodesList.Count; m++)
{
if (testingNodesList[m] == classifiedNodesList[m])
{
predictedCorrectValues += 1;
}
}
double accuracy = 0;
if (testingNodesList.Count != 0)
{
accuracy = (float)(predictedCorrectValues) / testingNodesList.Count * 100;
}
Console.Write("Found Accuracy:" + accuracy + " %");
Console.ReadKey();
}
public double set_gains(string inp_feature)
{
double sum_of_entropies_weighted_by_proportion = 0;
int s_dataset = Dataset.Rows.Count;
foreach (string unique_feature_value in Feature_uniqueValues[inp_feature])
{
DataTable temp_newds = getSubset(unique_feature_value, inp_feature);
int s_newds = temp_newds.Rows.Count;
double e_of_newds = 0.0;
double proportion = 0;
if (s_newds != 0)
{
proportion = (double)(s_newds) / s_dataset;
e_of_newds = StarterClass.get_entropy(temp_newds, null);
}
sum_of_entropies_weighted_by_proportion += proportion * e_of_newds;
}
double infogain = Entropy - sum_of_entropies_weighted_by_proportion;
return(infogain);
}