public Arff Parse(StreamReader reader)
{
string line = null;
Arff arff = new Arff ();
bool parsingData = false;
int attributeIdx = 0;
while ((line = reader.ReadLine()) != null) {
if (relation.IsMatch (line)) {
Match match = relation.Match (line);
arff.Relation = new Relation (match.Value);
continue;
}
if (attribute.IsMatch (line)) {
Match match = attribute.Match (line);
string name = match.Groups ["name"].Value;
List<Value> values = match
.Groups ["values"]
.Value
.Split (new char[] {','}, StringSplitOptions.None)
.ToList ()
.Select (x => new Value(x.Trim ()))
.ToList();
var att = new Attribute (attributeIdx, name, values);
arff.Attributes.Add (att);
arff.Target = att;
attributeIdx++;
continue;
}
if (data.IsMatch (line)) {
parsingData = true;
continue;
}
if (parsingData) {
arff.Data.Add (new Data (arff.Attributes, line));
continue;
}
}
if (arff.Relation == null || arff.Attributes.Count == 0 || arff.Data.Count == 0) {
throw new ParserException ();
}
// Remove the last attribute which is the Target attribute...
arff.Attributes.RemoveAt (arff.Attributes.Count - 1);
return arff;
}
public static void Main(string[] args)
{
if (args.Length > 2 || args.Length == 0)
{
Console.WriteLine("Bad usage. Use:\t./decision-tree.exe <arff file> \t[prune]");
Console.WriteLine("Example: \t./decision-tree.exe data/example.arff \tprune");
Environment.Exit(1);
}
bool prune = false;
string filename = args [0];
if (args.Length == 2)
{
prune = args [1].Equals("prune");
if (!prune)
{
Console.WriteLine("Pruning is disabled, could not parse \"prune\".");
}
}
if (!File.Exists(filename))
{
Console.WriteLine("File \"{0}\" does not exist.", filename);
Environment.Exit(1);
}
Console.WriteLine("File: \t\t{0}\nPruning: \t{1}", filename, prune);
ArffReader reader = new ArffReader();
Arff arff = reader.Parse(new StreamReader(filename));
Classifier c = new Classifier(arff, prune);
c.DrawTree();
}
public Classifier(Arff arff, bool prune = false)
{
this.arff = arff;
this.builder = new DecisionBuilder(this.arff);
this.prune = prune;
}
public DecisionBuilder(Arff arff)
{
this.arff = arff;
}
public Classifier(Arff arff, bool prune=false)
{
this.arff = arff;
this.builder = new DecisionBuilder (this.arff);
this.prune = prune;
}
public Arff Parse(StreamReader reader)
{
string line = null;
Arff arff = new Arff();
bool parsingData = false;
int attributeIdx = 0;
while ((line = reader.ReadLine()) != null)
{
if (relation.IsMatch(line))
{
Match match = relation.Match(line);
arff.Relation = new Relation(match.Value);
continue;
}
if (attribute.IsMatch(line))
{
Match match = attribute.Match(line);
string name = match.Groups ["name"].Value;
List <Value> values = match
.Groups ["values"]
.Value
.Split(new char[] { ',' }, StringSplitOptions.None)
.ToList()
.Select(x => new Value(x.Trim()))
.ToList();
var att = new Attribute(attributeIdx, name, values);
arff.Attributes.Add(att);
arff.Target = att;
attributeIdx++;
continue;
}
if (data.IsMatch(line))
{
parsingData = true;
continue;
}
if (parsingData)
{
arff.Data.Add(new Data(arff.Attributes, line));
continue;
}
}
if (arff.Relation == null || arff.Attributes.Count == 0 || arff.Data.Count == 0)
{
throw new ParserException();
}
// Remove the last attribute which is the Target attribute...
arff.Attributes.RemoveAt(arff.Attributes.Count - 1);
return(arff);
}
public DecisionBuilder(Arff arff)
{
this.arff = arff;
}