/// <summary>
/// Parses a random forest from a weka output file
/// </summary>
public static RandomForest FromWekaFile(string file)
{
var output = new RandomForest(new HashSet <string>(s_classLabels));
var reader = new StreamReader(file);
int id = 0;
string line;
while ((line = reader.ReadLine()) != null)
{
// keep reading until we find one of these. This one is followed
// by another line with equal signs and then empty lines until we find the first text
if (line.StartsWith("RandomTree"))
{
++id;
string firstLine;
reader.ReadLine();
while ((firstLine = reader.ReadLine()).Trim().Length == 0)
{
continue;
}
// now, we can start reading the tree
output.Trees.Add(RandomTree.FromStream(reader, firstLine, DefaultPrecision, id));
}
}
reader.Close();
return(output);
}
internal static RandomTree FromStream(StreamReader reader, string firstLine, int nodePrecision, int id)
{
var output = new RandomTree()
{
Id = id
};
var keepReading = true;
RandomTreeNode previous = null;
int initialId = 0;
bool first = true;
while (keepReading)
{
++initialId;
try
{
var line = first ? firstLine : reader.ReadLine();
var node = RandomTreeNode.BuildFromString(line, initialId, nodePrecision);
output.NumNodes = +1;
if (node.Level == 0)
{
// for roots
output.Roots.Add(node);
previous = node;
first = false;
}
else
{
// count the levels
if (node.Level > previous.Level)
{
// child
previous.Children.Add(node);
node.Parent = previous;
}
else if (node.Level < previous.Level)
{
// predecessor: travel backwards and look for the first parent with level = level
while (previous != null)
{
if (previous.Level == node.Level - 1)
{
// we found its parent
previous.Children.Add(node);
node.Parent = previous;
break;
}
previous = previous.Parent;
}
}
else
{
// same level, they are siblings
previous.Parent.Children.Add(node);
node.Parent = previous.Parent;
}
previous = node;
}
}
#pragma warning disable ERP022
catch (Exception)
{
// we could not read, so we are in the line after we finished.. how do we know? cause we could not build a predicate
break;
}
#pragma warning disable ERP022
}
// done...
return(output);
}