ErrorBasedPruning Class

public class ErrorBasedPruning

Public Class ErrorBasedPruning

// Suppose you have the following input and output data
// and would like to learn the relationship between the
// inputs and outputs by using a Decision Tree:

double[][] inputs = ...
int[] output = ...

// To prune a decision tree, we need to split your data into
// training and pruning groups. Let's say we have 100 samples,
// and would like to reserve 50 samples for training, and 50
// for pruning:

// Gather the first half for the training set
var trainingInputs = inputs.Submatrix(0, 49);
var trainingOutput = output.Submatrix(0, 49);

// Gather the second hand data for pruning
var pruningInputs = inputs.Submatrix(50, 99);
var pruningOutput = output.Submatrix(50, 99);


// Create the decision tree
DecisionTree tree = new DecisionTree( ... );

// Learn our tree using the training data
C45Learning c45 = new C45Learning(tree);
double error = c45.Run(trainingInputs, trainingOutput);


// Now we can attempt to prune the tree using the pruning groups
ErrorBasedPruning prune = new ErrorBasedPruning(tree, pruningInputs, pruningOutput);

// Gain threshold
prune.Threshold = 0.1;

double lastError;
double error = Double.PositiveInfinity;

do
{
    // Now we can start pruning the tree as 
    // long as the error doesn't increase

    lastError = error;
    error = prune.Run();

} while (error < lastError);