public DTestResults testTree(DataSet test)
{
int count = 0;
for(uint x=0; x < test.numRows(); x++)
{ count += (testRow(test.getRow(x))) ? 1 : 0; }
DTestResults retval = new DTestResults();
retval.correct = (uint) count;
retval.total = test.numRows();
return retval;
}
static int Main(string[] args)
{
if(args.Length != 2)
{
usage();
return 2;
}
//Console.WriteLine(new DataSet("sample1.txt"));
try
{
classify = new DataSet(args[0]);
test = new DataSet(args[1]);
}
catch(System.IO.IOException e)
{ Console.WriteLine(e.Message); }
//make tree
DTree tree = new DTree(classify);
tree.printTree();
DTestResults results = tree.testTree(test);
Console.WriteLine("Correct/Total: " + results.correct + " / " + results.total);
Console.WriteLine("Error Rate: " + ((double)results.correct/(double)results.total));
//Console.WriteLine(classify.gain(3));
//Console.WriteLine(classify.entropyAttr(3,"Weak"));
//Console.WriteLine(classify.entropy);
//Console.WriteLine(classify.partitionData(0,"Rain"));
//Console.WriteLine("Num Attribs: " + classify.numAttributes());
/*
bool[] marked = new bool[classify.numAttributes()];
for(int x=0; x < marked.Length; x++)
{ marked[x] = false; }
for(int x=0; x < marked.Length; x++)
{
uint a = classify.bestGain(marked);
marked[a] = true;
Console.WriteLine(a);
}
Console.WriteLine(classify.getMostOccuringClassification());
//Console.WriteLine("A" + classify.gain(2));
*/
return 0;
}
public DataSet partitionData(uint aidx, string v)
{
DataSet retval = new DataSet(attribs.getLength());
for(uint x=0; x < size; x++)
{
if(rows[x].hasAttribute(aidx,v))
{ retval.addRow(rows[x].originalString()); }
}
return retval;
}
public DNode(DataSet classify, ref bool[] marked, string path)
{
//Console.WriteLine(markedString(marked));
classification = null; //only gets set on leaf node
attr_assigned = false; //unassigned
this.path = path;
//if the tree is perfectly classified (entropy == 0)
// or if we're out of attributes, return the most occuring one
if(classify.entropy == 0 || allMarked(ref marked))
{ classification = classify.getMostOccuringClassification();}
else
{
attribute = classify.bestGain(marked);
attr_assigned = true;
children = new DNode[classify.getAttributes(attribute).Length];
//Console.WriteLine("Attrib: " + attribute + " Children: " + classify.getAttributes(attribute).Length);
for(uint x=0; x < children.Length; x++)
{
marked[attribute] = true;
string tmp = classify.getAttributes(attribute)[x];
children[x] = new DNode(classify.partitionData(attribute,tmp),ref marked,tmp);
}
}
}
public DTree(DataSet classify)
{
bool[] marked = new bool[classify.numAttributes()];
root = new DNode(classify,ref marked, "");
}