baseCode.math
Class ROC

java.lang.Object
  baseCode.math.ROC

public class ROC

extends java.lang.Object

Functions for calculating Receiver operator characteristics.

Copyright (c) 2004 Columbia University

Version:

$Id: ROC.java,v 1.13 2004/08/16 00:21:40 pavlidis Exp $

Author:

Paul Pavlidis

Constructor Summary

ROC()

Method Summary

static double	aroc(int totalSize, java.util.Set ranks) Calculate area under ROC.
static double	aroc(int totalSize, java.util.Set ranks, int maxFP) Calculate area under ROC, up to a given number of False positives.
static double	rocpval(int numpos, double aroc) For an AROC value, calculates a p value based on approximation for calculating the stanadard deviation.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

ROC

public ROC()

Method Detail

aroc

public static double aroc(int totalSize,
                          java.util.Set ranks)

Calculate area under ROC. The input is the total number of items in the data, and the ranks of the positives in the current ranking. LOW ranks are considered better. (e.g., rank 0 is the 'best')

Parameters:

totalSize - int

ranks - Map

Returns:

AROC

aroc

public static double aroc(int totalSize,
                          java.util.Set ranks,
                          int maxFP)

Calculate area under ROC, up to a given number of False positives. The input is the total number of items in the data, and the ranks of the positives in the current ranking. LOW ranks are considered better. (e.g., rank 0 is the 'best')

Parameters:

totalSize - int

ranks - Map

maxFP - - the maximum number of false positives to see before stopping. Set to 50 to get the Gribskov roc50. If maxFP <= 0, it is ignored.

Returns:

AROC

rocpval

public static double rocpval(int numpos,
                             double aroc)

For an AROC value, calculates a p value based on approximation for calculating the stanadard deviation. Highly approximate!

Parameters:

numpos - How many positives are in the data.

aroc - The AROC

Returns:

The p value.