public static XgbModel GetXgbModelFromFile(string fileName, int maxTrees)
{
XgbModel xgbModel = new XgbModel();
if (!File.Exists(fileName))
{
Console.ForegroundColor = ConsoleColor.Yellow;
Console.WriteLine(String.Format("Error: File {0} does not exist.",fileName));
Console.ResetColor();
return null;
}
Console.ResetColor();
Console.WriteLine(String.Format("Parsing {0}", fileName));
int numTree = 0;
var fileInfo = new FileInfo(fileName);
GlobalStats.ModelFileSize = fileInfo.Length;
using (StreamReader sr = new StreamReader(fileName))
{
while(!sr.EndOfStream) {
var line = sr.ReadLine().Trim();
if (line.StartsWith("booster") || line == String.Empty)
{
if (xgbNodeList.Count > 0)
{
numTree++;
Console.ForegroundColor = ConsoleColor.DarkGreen;
Console.Write(String.Format("Constructing tree #{0} ", numTree));
XgbTree tree = new XgbTree(xgbNodeList[0]);
ConstructXgbTree(tree);
Console.WriteLine(String.Format("=> depth: {0} ({1} nodes)", tree.Height, xgbNodeList.Count));
Console.ResetColor();
xgbModel.XgbTrees.Add(tree);
xgbNodeList.Clear();
if (numTree == maxTrees) break;
}
}
else
{
var node = ParseXgbTreeNode(line);
if (node == null)
{
return null;
}
xgbNodeList.Add(node.Number, node);
}
}
}
if (xgbNodeList.Count > 0 && (maxTrees < 0 || numTree < maxTrees))
{
numTree++;
Console.ForegroundColor = ConsoleColor.DarkGreen;
Console.Write(String.Format("Constructing tree #{0} ", numTree));
XgbTree tree = new XgbTree(xgbNodeList[0]);
ConstructXgbTree(tree);
Console.WriteLine(String.Format("=> depth: {0} ({1} nodes)", tree.Height, xgbNodeList.Count));
Console.ResetColor();
xgbModel.XgbTrees.Add(tree);
xgbNodeList.Clear();
}
Console.ForegroundColor = ConsoleColor.Green;
Console.WriteLine(String.Format("{0} trees has been constructed.\n", xgbModel.NumTrees));
Console.ResetColor();
return xgbModel;
}
private void CollectFeatureInteractions(XgbTree tree, HashSet<XgbTreeNode> currentInteraction, double currentGain, double currentCover, double pathProbability, int depth, int deepening)
{
if (tree.IsLeafNode)
{
return;
}
currentInteraction.Add(tree.Data);
currentGain += tree.Data.Gain;
currentCover += tree.Data.Cover;
var pathProbabilityLeft = pathProbability * (((XgbTree)tree.Left).Data.Cover / tree.Data.Cover);
var pathProbabilityRight = pathProbability * (((XgbTree)tree.Right).Data.Cover / tree.Data.Cover);
var fi = new FeatureInteraction(currentInteraction, currentGain, currentCover, pathProbability, depth, _treeIndex, 1);
if (depth < _maxDeepening || _maxDeepening < 0)
{
var newInteractionLeft = new HashSet<XgbTreeNode>() { };
var newInteractionRight = new HashSet<XgbTreeNode>() { };
CollectFeatureInteractions((XgbTree)tree.Left, newInteractionLeft, 0, 0, pathProbabilityLeft, depth + 1, deepening + 1);
CollectFeatureInteractions((XgbTree)tree.Right, newInteractionRight, 0, 0, pathProbabilityRight, depth + 1, deepening + 1);
}
var path = string.Join("-", currentInteraction.Select(x => x.Number));
if (!_treeFeatureInteractions.ContainsKey(fi.Name))
{
_treeFeatureInteractions.Add(fi.Name, fi);
_pathMemo.Add(path);
}
else
{
if(_pathMemo.Contains(path))
{
return;
}
_pathMemo.Add(path);
var tfi = _treeFeatureInteractions[fi.Name];
tfi.Gain += currentGain;
tfi.Cover += currentCover;
tfi.FScore += 1;
tfi.FScoreWeighted += pathProbability;
tfi.AverageFScoreWeighted = tfi.FScoreWeighted / tfi.FScore;
tfi.AverageGain = tfi.Gain / tfi.FScore;
tfi.ExpectedGain += currentGain * pathProbability;
tfi.TreeDepth += depth;
tfi.AverageTreeDepth = tfi.TreeDepth / tfi.FScore;
tfi.TreeIndex += _treeIndex;
tfi.AverageTreeIndex = tfi.TreeIndex / tfi.FScore;
tfi.SplitValueHistogram.Merge(fi.SplitValueHistogram);
}
if (currentInteraction.Count - 1 == _maxInteractionDepth)
return;
var currentInteractionLeft = new HashSet<XgbTreeNode>(currentInteraction);
var currentInteractionRight = new HashSet<XgbTreeNode>(currentInteraction);
var leftTree = (XgbTree)(tree.Left);
var rightTree = (XgbTree)(tree.Right);
if (leftTree.IsLeafNode && deepening == 0)
{
var tfi = _treeFeatureInteractions[fi.Name];
tfi.SumLeafValuesLeft += leftTree.Data.LeafValue;
tfi.SumLeafCoversLeft += leftTree.Data.Cover;
tfi.HasLeafStatistics = true;
}
if (rightTree.IsLeafNode && deepening == 0)
{
var tfi = _treeFeatureInteractions[fi.Name];
tfi.SumLeafValuesRight += rightTree.Data.LeafValue;
tfi.SumLeafCoversRight += rightTree.Data.Cover;
tfi.HasLeafStatistics = true;
}
CollectFeatureInteractions((XgbTree)tree.Left, currentInteractionLeft, currentGain, currentCover, pathProbabilityLeft, depth + 1, deepening);
CollectFeatureInteractions((XgbTree)tree.Right, currentInteractionRight, currentGain, currentCover, pathProbabilityRight, depth + 1, deepening);
}
private static void ConstructXgbTree(XgbTree tree)
{
if (tree.Data.LeftChild != null)
{
tree.Add(new XgbTree(xgbNodeList[(int)tree.Data.LeftChild]));
ConstructXgbTree((XgbTree)tree.Left);
}
if (tree.Data.RightChild != null)
{
tree.Add(new XgbTree(xgbNodeList[(int)tree.Data.RightChild]));
ConstructXgbTree((XgbTree)tree.Right);
}
}
private void CollectFeatureInteractions(XgbTree tree, HashSet<XgbTreeNode> currentInteraction, double currentGain, double currentCover, double pathProbability, int depth)
{
if (tree.IsLeafNode)
return;
currentInteraction.Add(tree.Data);
currentGain += tree.Data.Gain;
currentCover += tree.Data.Cover;
var pathProbabilityLeft = pathProbability * (((XgbTree)tree.Left).Data.Cover / tree.Data.Cover);
var pathProbabilityRight = pathProbability * (((XgbTree)tree.Right).Data.Cover / tree.Data.Cover);
var fi = new FeatureInteraction(currentInteraction, currentGain, currentCover, pathProbability, 1);
if (depth < _maxDeepening || _maxDeepening < 0)
{
var newInteractionLeft = new HashSet<XgbTreeNode>() { };
var newInteractionRight = new HashSet<XgbTreeNode>() { };
CollectFeatureInteractions((XgbTree)tree.Left, newInteractionLeft, 0, 0, pathProbabilityLeft, depth + 1);
CollectFeatureInteractions((XgbTree)tree.Right, newInteractionRight, 0, 0, pathProbabilityRight, depth + 1);
}
var path = string.Join("-", currentInteraction.Select(x => x.Number));
if (!_treeFeatureInteractions.ContainsKey(fi.Name))
{
_treeFeatureInteractions.Add(fi.Name, fi);
_pathMemo.Add(path);
}
else
{
// reoccurrence?
if (!_pathMemo.Contains(path))
{
_pathMemo.Add(path);
var tfi = _treeFeatureInteractions[fi.Name];
tfi.Gain += currentGain;
tfi.Cover += currentCover;
tfi.FScore += 1;
tfi.FScoreWeighted += pathProbability;
tfi.AverageFScoreWeighted = tfi.FScoreWeighted / tfi.FScore;
tfi.AverageGain = tfi.Gain / tfi.FScore;
tfi.ExpectedGain += currentGain * pathProbability;
}
return;
}
if (currentInteraction.Count - 1 == _maxInteractionDepth)
return;
var currentInteractionLeft = new HashSet<XgbTreeNode>(currentInteraction);
var currentInteractionRight = new HashSet<XgbTreeNode>(currentInteraction);
CollectFeatureInteractions((XgbTree)tree.Left, currentInteractionLeft, currentGain, currentCover, pathProbabilityLeft, depth + 1);
CollectFeatureInteractions((XgbTree)tree.Right, currentInteractionRight, currentGain, currentCover, pathProbabilityRight, depth + 1);
}