public Frequency(AttributePack ap, string attributeName)
{
string classifierName = ap.classifierName;
attributePack = ap;
foreach (AttributeSet ds in ap.attributeSets)
{
//Values durchgehen
string attributeValue;
ds.attributes.TryGetValue(attributeName, out attributeValue);
// check ob bereits existiert
if (!_freqTable.ContainsKey(attributeValue))
{
_freqTable.Add(attributeValue, new Dictionary <string, int>());
}
// classifier bestimmen
string classifierValue;
ds.attributes.TryGetValue(classifierName, out classifierValue);
// classifier value checken und hochzählen
Dictionary <string, int> a;
if (_freqTable.TryGetValue(attributeValue, out a))
{
if (!a.ContainsKey(classifierValue))
{
a.Add(classifierValue, 0);
}
}
_freqTable[attributeValue][classifierValue]++;
}
}
public void buildTree(AttributePack dp, TreeNode node)
{
double maxGain = 0;
string maxAttribute = null;
if (dp.attributeNames.Count < 1)
{
//leave node mit mehrheit
node.value = dp.attributeSets.Max(ds => ds.attributes[dp.classifierName]);
return;
}
if (dp.attributeSets.All(ds => ds.attributes[dp.classifierName] == dp.attributeSets.First().attributes[dp.classifierName]))
{
node.value = dp.attributeSets.First().attributes[dp.classifierName];
return;
}
foreach (string attributeName in dp.attributeNames)
{
// create Frequency table
Frequency table = new Frequency(dp, attributeName);
double gain = table.calcInfoGain();
if (gain > maxGain)
{
maxGain = gain;
maxAttribute = attributeName;
}
}
node.value = maxAttribute;
Dictionary <string, AttributePack> dataPacks = dp.sortDatapack(maxAttribute);
foreach (var d in dataPacks)
{
TreeNode newNode = new TreeNode();
node.nodes.Add(d.Key, newNode);
buildTree(d.Value, newNode);
}
}
