RandomForest Class Reference

RandomForest class. Interface for growing random forests from training data or loading a random forest from disk.

#include <random_forest.h>

List of all members.

Public Member Functions
	RandomForest ()
	Empty constructor.
	RandomForest (const InstanceSet &set, int num_trees, int K, int max_depth)
	Constructor. (Build from training data).
int	predict (const InstanceSet &set, int instance_no) const
	Method to predict the label.
float	predict_prob (const InstanceSet &set, int instance_no, int label) const
	Predict probability of given label.
float	testing_accuracy (const InstanceSet &testset) const
	Returns test accuracy of a labeled test set.
float	training_accuracy () const
	Returns training accuracy.
float	oob_accuracy () const
	Returns OOB accuracy (unbiased estimate of test accuracy).
void	variable_importance (vector< pair< float, int > > *ranking, unsigned int *seed) const
	Variable importance ranking of features.
void	read (istream &i)
	Load random forest.
void	write (ostream &o)
	Save random forest.
void	print () const
	Debug output.

---

Constructor & Destructor Documentation

RandomForest (  const InstanceSet &  set, int  num_trees, int  K, int  max_depth )

Parameters: set  training data num_trees  #trees to train K  #random vars to consider at each split max_depth  this param should be deprecated (depth is limited by number of instances)

---

The documentation for this class was generated from the following files:

• random_forest.h
• random_forest.cc

---