internal C45FeatureSetCollection GetFSCFromFile(String filename)
{
C45FeatureSetCollection featureSetCollection = new C45FeatureSetCollection();
C45Schema schema = C45Schema.CreateFromFile(filename.Replace(".data".ToString(), ".names".ToString()));
int character = 0;
String line = String.Empty;
FileStream file = new FileStream(
filename,
FileMode.Open);
while (character >= 0)
{
line = String.Empty;
character = file.ReadByte();
while (character != 0x0A &&
character > 0)
{
line += Convert.ToChar(character);
character = file.ReadByte();
}
featureSetCollection.Add(C45FeatureSet.GetFSFromLine(line, schema));
}
return(featureSetCollection);
}
private void AppendFeatureValueFromGridView(
DataGridViewCell cell,
C45FeatureSet featureSet)
{
if (cell.ColumnIndex < this.schema.Properties.Count)
{
String cellValue = String.Empty;
if (cell.IsInEditMode)
{
cellValue = cell.EditedFormattedValue.ToString();
}
else
{
cellValue = cell.FormattedValue.ToString();
}
C45SchemaProperty col = (C45SchemaProperty)this.dataGridView.Columns[cell.ColumnIndex].Tag;
if (cellValue == "?".ToString() ||
cellValue == "".ToString())
{
featureSet.AddUnknownFeature();
}
else
{
if (col.IsDiscrete)
{
// Just one attribute with 100 % probability supported, yet.
// Multiple attribute has to be implemented in future.
C45DiscreteFeature discreteFeature = new C45DiscreteFeature();
C45DiscreteFeatureAttribute discreteFeatureAttribute = new C45DiscreteFeatureAttribute();
discreteFeatureAttribute.AttributeName = (String)cellValue;
discreteFeatureAttribute.Probability = 100;
discreteFeature.Add(discreteFeatureAttribute);
featureSet.AddDiscreteFeature(discreteFeature);
}
else
{
// Just a fix continous value or ignore-value supported, yet.
// Lower and upper bound has to be implemented in future.
C45ContinuousFeature continuousFeature = new C45ContinuousFeature();
continuousFeature.LowerBound = Convert.ToInt32(cellValue);
continuousFeature.UpperBound = Convert.ToInt32(cellValue);
featureSet.AddContinousFeature(continuousFeature);
}
}
}
}
private C45FeatureSet GetFeatureSetFromGrid(DataGridViewRow row)
{
C45FeatureSet newSet = new C45FeatureSet();
foreach (DataGridViewCell cell in row.Cells)
{
this.AppendFeatureValueFromGridView(cell, newSet);
}
newSet.IsMarked = row.Selected;
newSet.UserResult.ClassName = row.Cells[this.userClassColumn].FormattedValue.ToString();
return(newSet);
}
internal static C45FeatureSet GetFSFromLine(String line, C45Schema schema)
{
C45FeatureSet featureSet = new C45FeatureSet();
int currentFeature = 0;
String[] features = line.Split(",".ToCharArray());
for (int i = 0; i < features.Length; i++)
{
features[i] = features[i].Trim();
}
foreach (C45SchemaProperty column in schema.Properties)
{
if (features[currentFeature] == "" ||
features[currentFeature] == "?")
{
featureSet.AddUnknownFeature();
}
else
{
if (column.IsDiscrete)
{
C45DiscreteFeature discreteFeature = new C45DiscreteFeature();
C45DiscreteFeatureAttribute discreteFeatureAttribute = new C45DiscreteFeatureAttribute();
discreteFeatureAttribute.AttributeName = features[currentFeature];
discreteFeatureAttribute.Probability = 100;
discreteFeature.Add(discreteFeatureAttribute);
featureSet.AddDiscreteFeature(discreteFeature);
}
else
{
C45ContinuousFeature continuousFeature = new C45ContinuousFeature();
continuousFeature.LowerBound = Convert.ToInt32(features[currentFeature]);
continuousFeature.UpperBound = Convert.ToInt32(features[currentFeature]);
featureSet.AddContinousFeature(continuousFeature);
}
}
currentFeature++;
}
featureSet.UserResult.ClassName = features[currentFeature];
return(featureSet);
}
private void InsertFeatureSetIntoGridView(
C45FeatureSet fs,
int rowIndex)
{
int columnIndex = 0;
foreach (object f in fs)
{
this.InsertFeatureValueIntoGridView(
f,
columnIndex,
rowIndex);
columnIndex++;
}
}
public void AddResult(
C45FeatureSet featureSet,
String result)
{
try
{
String dataFile = this.Settings.C45SchemaFilename + @".data";
FileStream fs = new FileStream(dataFile,
FileMode.Append);
String newCase = String.Empty;
for (int i = 0; i < featureSet.Count; i++)
{
if (featureSet._featureArray[i].bIsKnown)
{
if (featureSet._featureArray[i].bIsDiscreteFeature)
{
newCase += this.ConvertByteArrayToString(featureSet._featureArray[i].discreteFeature.data, 4) + @",";
}
else
{
//Calculate the average continuous value
int continuous = Convert.ToInt32((featureSet._featureArray[i].continuousFeature.fLowerBound + featureSet._featureArray[i].continuousFeature.fUpperBound) / 2);
newCase += continuous.ToString() + @",";
}
}
else
{
newCase += @"?,";
}
}
newCase += result;
fs.WriteByte(0x0A);
foreach (byte b in newCase.ToCharArray())
{
fs.WriteByte(b);
}
fs.Close();
}
catch (Exception ex)
{
throw (new C45Exception(C45Exceptions.CASES_NOT_ADDED, ex));
}
}