static void ExampleML4()
{
string file_path_ser = @"D:\Dropbox\TestData\ML4\brains_dataset.ser";
IDataSetHybrid data_set_hybrid_unlabeled = null;
if (File.Exists(file_path_ser))
{
try
{
data_set_hybrid_unlabeled = DataSetHybrid.Read(new BinaryReader(File.Open(file_path_ser, FileMode.Open)));
}
catch (Exception)
{
data_set_hybrid_unlabeled = null;
}
}
if (data_set_hybrid_unlabeled == null)
{
IDataSet <int> data_set0 = ReadExcellNominal(@"D:\Dropbox\TestData\ML4\brains_ml.xlsx");
data_set_hybrid_unlabeled = CleanUpML4(data_set0);
ToolsIOSerialization.SerializeToFile(file_path_ser, data_set_hybrid_unlabeled);
}
IDataSet <int> data_set_nominal_unlabeled = data_set_hybrid_unlabeled.ConvertToNominal(5);
IDataSet <int, int> data_set_nominal_labeled = data_set_nominal_unlabeled.PromoteFeatureToLabel(0);
Tuple <IDataSet <int, int>, IDataSet <int, int> > split = data_set_nominal_labeled.Split();
IDataSet <int, int> training_set = split.Item1;
IDataSet <int, int> test_set = split.Item2;
ProccessDataSetNominalJoinTable(training_set, test_set);
}
public static IDataSetHybrid CleanUpML4(IDataSet <int> data_set0)
{
// Manipulate it
Console.WriteLine("For feature index: " + 0 + " Adding value type as missing: " + "9");
Console.WriteLine("Removing all instances missing: " + data_set0.DataContext.GetFeatureName(0));
IDataSet <int> data_set1 = data_set0.RemoveInstancesMissing(0);
Console.WriteLine("Removing all instances missing: " + data_set0.DataContext.GetFeatureName(1));
IDataSet <int> data_set2 = data_set1.RemoveInstancesMissing(1);
int[] remove_feature_indexes = new int[] { 1, 2, 3, 5, 7, 8, 9, 11, 12 };
foreach (int remove_feature_index in remove_feature_indexes)
{
Console.WriteLine("Removing: " + data_set2.DataContext.GetFeatureName(remove_feature_index));
}
IDataSet <int> data_set3 = data_set2.RemoveFeatures(remove_feature_indexes);
int[] interval_feature_indexes = new int[] { 1, 2, 3 };
foreach (int interval_feature_index in interval_feature_indexes)
{
Console.WriteLine("Promoting to hybrid: " + data_set0.DataContext.GetFeatureName(interval_feature_index));
}
IDataSetHybrid data_set4 = data_set3.PromoteFeatureLevelToInterval(interval_feature_indexes);
return(data_set4);
}