Tree runID3(IEnumerable<XElement> examples, XName target_attribute
, List<XName> attribute_list, string exp) {
Tree root = new Tree();
if (Entropy(examples) == 0.0) {
if (getPercentPositive() == 1.0) {
_expressions.Add("(" + exp.Remove(0, 3) + ")");
root.setAttributeName("Yes");
} else root.setAttributeName("No");
return root;
}
if (attribute_list.Count == 0) {
//only target attribute exists
if (getPercentPositive() >= 0.5) {
_expressions.Add("(" + exp.Remove(0, 3) + ")");
root.setAttributeName("Yes");
} else root.setAttributeName("No");
return root;
}
//now do main work of algorithm
XName best = getBestAttribute(examples, attribute_list);
List<string> p = _dataSet.getPossibleAttributeValues(best);
root.setAttributeName(best.ToString());
attribute_list = removeBestAttribute(attribute_list, best);
exp += " ^ " + best.ToString() + " = ";
foreach (string s in p) {
IEnumerable<XElement> subset = _dataSet.refineQuery(examples, best, s);
Tree t = runID3(subset, target_attribute, attribute_list, (exp + s));
t.setAttributeValue(s);
root.Nodes.Add(t);
}
return root;
}
