static void classify(TextWriter writer, char csv_seperator, char dec_seperator)
{
ModelLoader loader = new ModelLoader();
DataController datacontroller = new DataController(csv_seperator, dec_seperator);
// Check if the data is already classified, and thus we are in verification mode and will check how our model holds up.
string verification_mode_string = writer.askFromConfig("Verification mode? (TRUE or FALSE) ", "CLASSIFICATION", "verification-mode");
bool verification_mode = (verification_mode_string == "TRUE");
// Import the classification data.
string data_location = writer.askFromConfig("Enter the file path to import the data from. ", "CLASSIFICATION", "input-location");
ObservationSet observations = verification_mode ? datacontroller.importExamples(data_location) : datacontroller.importUnclassified(data_location);
string classifier_name = observations.target_attribute;
// Import the model
string model_location = writer.askFromConfig("Enter the file path to import the model from. ", "CLASSIFICATION", "model-location");
DecisionTree model = loader.load_model(model_location, classifier_name);
// Get ready for classification.
string export_location = writer.askFromConfig("Enter the file path to export data to. ", "CLASSIFICATION", "output-location");
Console.WriteLine($"READY for classification{(verification_mode ? " in verification mode" : "")}. Press a key to start classification process \n");
Console.ReadKey(true);
List <DataInstance> classified_instances = new List <DataInstance>();
int correct_classifications = 0;
foreach (DataInstance instance in observations.instances)
{
string prediction = model.classify(instance);
if (verification_mode)
{
if (instance.getProperty(classifier_name) == prediction)
{
correct_classifications++;
}
}
else
{
instance.setProperty(classifier_name, model.classify(instance));
}
classified_instances.Add(instance);
}
ObservationSet export_set = new ObservationSet(classified_instances, classifier_name, observations.attributes);
Console.WriteLine("Classification succesful, saving now.");
if (verification_mode)
{
double succesPercentage = Math.Round(((double)correct_classifications / (double)classified_instances.Count) * 100, 2);
Console.WriteLine($"Verification mode results: success rate of {succesPercentage}% ({correct_classifications} / {classified_instances.Count}).");
}
datacontroller.exportSet(export_location, export_set);
}
public void Make(string name, DecisionTree tree)
{
long beforeSnapShot = this.stopwatch.ElapsedMilliseconds;
// Save model
writer.set_location(location);
writer.filesave_lines(name + "." + this.model_extension, ModelManager.generate_model(tree));
// Save model as rule set
writer.filesave_lines(name + "." + this.rules_extension, ModelManager.generate_rules(tree));
// Save image
writer.filesave_lines(name + "." + this.drawing_extension, drawer.lines(tree));
// Lets see how well it classifies.
List <DataInstance> total_set = new List <DataInstance>();
foreach (Leaf leaf in tree.data_locations.Keys.ToList())
{
total_set.AddRange(tree.data_locations[leaf]);
}
int correct_classifications = 0;
foreach (DataInstance instance in total_set)
{
if (tree.classify(instance) == instance.getProperty(tree.target_attribute))
{
correct_classifications++;
}
}
double succesPercentage = Math.Round(((double)correct_classifications / (double)total_set.Count) * 100, 2);
Console.WriteLine($"[SNAPSHOT] : Snapshot {name} made with success rate of {succesPercentage}% ({correct_classifications} / {total_set.Count}).");
this.secondsBySnapShot += (this.stopwatch.ElapsedMilliseconds - beforeSnapShot);
}