weka.classifiers

Interface IntervalEstimator

All Known Implementing Classes:

GaussianProcesses

public interface IntervalEstimator

Interface for classifiers that can output confidence intervals

Version:

$Revision: 1.2 $

Author:

Kurt Driessens (kurtd@cs.waikato.ac.nz)

Method Summary

Modifier and Type	Method and Description
double[][]	predictInterval(Instance inst, double confidenceLevel) Returns an N*2 array, where N is the number of possible classes, that estimate the boundaries for the confidence interval with a confidence level specified by the second parameter.

Method Detail

predictInterval

double[][] predictInterval(Instance inst,
                           double confidenceLevel)
                    throws java.lang.Exception

Returns an N*2 array, where N is the number of possible classes, that estimate the boundaries for the confidence interval with a confidence level specified by the second parameter. Every row of the returned array gives the probability estimates for a single class. In the case of numeric predictions, a single confidance interval will be returned.

Parameters:

inst - the instance to make the prediction for.

confidenceLevel - the percentage of cases that the interval should cover.

Returns:

an array of confidance intervals (one for each class)

Throws:

java.lang.Exception - if the intervals can't be computed