TMVA::DecisionTree

 default constructor using the GiniIndex as separation criterion, 
 no restrictions on minium number of events in a leave note or the
 separation gain in the node splitting

 default constructor using the GiniIndex as separation criterion, 
 no restrictions on minium number of events in a leave note or the
 separation gain in the node splitting

 constructor specifying the separation type, the min number of
 events in a no that is still subjected to further splitting, the
 number of bins in the grid used in applying the cut for the node
 splitting.

 copy constructor that creates a true copy, i.e. a completely independent tree 
 the node copy will recursively copy all the nodes 

 destructor
 too bad, I cannot delete this as I don't know how to make a proper copy
 when copying a decsiont Tree.. hence so far I just copy the pointer, which
 means I cannot delet it here...
   delete fQualityIndex; 

 descend a tree to find all its leaf nodes, fill max depth reached in the
 tree at the same time. 

 building the decision tree by recursively calling the splitting of 
 one (root-) node into two daughter nodes (returns the number of nodes)

 fill the existing the decision tree structure by filling event
 in from the top node and see where they happen to end up

 fill the existing the decision tree structure by filling event
 in from the top node and see where they happen to end up

 clear the tree nodes (their S/N, Nevents etc), just keep the structure of the tree

 prune (get rid of internal nodes) the Decision tree to avoid overtraining
 serveral different pruning methods can be applied as selected by the 
 variable "fPruneMethod". Currently however only the Expected Error Pruning
 is implemented
 

 recursive prunig of nodes using the Expected Error Pruning (EEP)
 if internal node, then prune

 prunig of nodes using the Cost Complexity criteria. The Pruning is performed
 until a minimum in the cost complexity CC(alpha) is reached.
 CC(alpha) = alpha*NLeafs + sum_over_leafs[ N*Quality(leaf) ]
 where Quality(leaf) is given by the 1-purity (for Misclassification Error)
 purity(1-purity) for Gini-Index..e.t.c.. typically the Misclassification Error
 is used for guiding the pruning.

 Similar to the CostCoplexity pruning, only here I calculate immediately
 the "prunestrength" (= alpha, the regularisation parameter in the CostComplexity)
 for which the respective subtree below a node would be pruned. Then I continue
 pruning until all nodes have such a value larger than the specified prunestrength

 get the weakest link for the pruning step

 loop over all non-leaf nodes of the tree and calculate for each
 of these nodes the prunestrenght ("alpha") at which this node
 would win against its subtree.(hence it's subtree would be pruned)
 this is given by:
          R(t) - R(T_t)       where R(t)  : MisClassificationCost of the node t
 alpha < --------------             R(T_t): MisClassificationCost of subtree blow t
          |T_t| - 1                 |T|   : # nodes in Tree T

 get the misclassificationCost of the subTree

 get the misclassificationCost of the subTree

 return the number of terminal nodes in the sub-tree below Node n

 returns the cost complexity criterion for the decision tree
 see "L.Breiman, J.H.Friedman, R.A.Olshen, C.J.Stone; "Classification and
 Regression Trees", Wadsworth International Group (1984), Chapman & Hall/CRC (1984)

 returns the cost complexity criterion for the decision tree
 see "L.Breiman, J.H.Friedman, R.A.Olshen, C.J.Stone; "Classification and
 Regression Trees", Wadsworth International Group (1984), Chapman & Hall/CRC (1984)

 traverse the whole tree and fill the map of QualityGain - Node
 for pre-leaf nodes, deciding which node is the next prune candidate

 traverse the whole tree and find the leaf nodes, and then fill the Quality Criterion
 for the leaf nodes (Used in the Pruning)

 descend a tree to find all its leaf nodes

 get the  pruning candidate node as the one pre-leaf node that gives
 the smalest quality gain

 prune away the subtree below the node 

 calculate an UPPER limit on the error made by the classification done
 by this node. If the S/S+B of the node is f, then according to the
 training sample, the error rate (fraction of misclassified events by
 this node) is (1-f)
 now f has a statistical error according to the binomial distribution
 hence the error on f can be estimated (same error as the binomial error
 for efficency calculations ( sigma = sqrt(eff(1-eff)/nEvts ) ) 

 calculate the expected statistical error on the subtree below "node"
 which is used in the expected error pruning

 get left daughter node current node "n"

 get right daughter node current node "n"

 retrieve node from the tree. Its position (up to a maximal tree depth of 64)
 is coded as a sequence of left-right moves starting from the root, coded as
 0-1 bit patterns stored in the "long-integer"  (i.e. 0:left ; 1:right

 helper function which calculates gets the Min and Max value
 of the event variables in the current event sample

 helper function which calculates the grid points used for
 the cut scan

 decide how to split a node. At each node, ONE of the variables is
 choosen, which gives the best separation between signal and bkg on
 the sample which enters the Node.  
 In order to do this, for each variable a scan of the different cut
 values in a grid (grid = fNCuts) is performed and the resulting separation
 gains are compared.. This cut scan uses either a binary search tree
 or a simple loop over the events depending on the number of events
 in the sample 

 the event e is put into the decision tree (starting at the root node)
 and the output is NodeType (signal) or (background) of the final node (basket)
 in which the given events ends up. I.e. the result of the classification if
 the event for this decision tree.

calculates the purity S/(S+B) of a given event sample

return the relative variable importance, normalized to all
variables together having the importance 1. The importance in
evaluated as the total separation-gain that this variable had in
the decision trees (weighted by the number of events)

 returns the relative improtance of variable ivar

 returns poiter to 2D histogram with tree drawing in it.

 recursively puts an entries in the histogram for the node and its daughters