/*
* Copyright (C) 2015 Edward Raff
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see .
*/
package jsat.distributions.discrete;
import static jsat.math.SpecialMath.*;
import static java.lang.Math.*;
/**
* The Binomial distribution is the distribution for the number of successful,
* independent, trials with a specific probability of success
*
* @author Edward Raff
*/
public class Binomial extends DiscreteDistribution
{
private int trials;
private double p;
/**
* Creates a new Binomial distribution for 1 trial with a 0.5 probability of
* success
*/
public Binomial()
{
this(1, 0.5);
}
/**
* Creates a new Binomial distribution
*
* @param trials the number of independent trials
* @param p the probability of success
*/
public Binomial(int trials, double p)
{
setTrials(trials);
setP(p);
}
/**
* The number of trials for the distribution
* @param trials the number of trials to perform
*/
public void setTrials(int trials)
{
if(trials < 1)
throw new IllegalArgumentException("number of trials must be positive, not " + trials);
this.trials = trials;
}
public int getTrials()
{
return trials;
}
/**
* Sets the probability of a trial being a success
* @param p the probability of success for each trial
*/
public void setP(double p)
{
if(Double.isNaN(p) || p < 0 || p > 1)
throw new IllegalArgumentException("probability of success must be in [0, 1], not " + p);
this.p = p;
}
public double getP()
{
return p;
}
@Override
public double logPmf(int x)
{
if(x > trials || x < 0)
return -Double.MAX_VALUE;
//re write as: log((Gamma(n+1) p^x (1-p)^(n-x))/(Gamma(x+1) Gamma(n-x+1)))
//then expand to: n log(1-p)-log(Gamma(n-x+1))+log(Gamma(n+1))-x log(1-p)+x log(p)-log(Gamma(x+1))
final int n = trials;
return n*log(1-p) - lnGamma(n-x+1) + lnGamma(n+1) - x*log(1-p)+ x * log(p) - lnGamma(x+1);
}
@Override
public double pmf(int x)
{
if(x > trials || x < 0)
return 0;
return exp(logPmf(x));
}
@Override
public double cdf(int x)
{
if(x >= trials)
return 1;
if(x < 0)
return 0;
return betaIncReg(1-p, trials-x, 1+x);
}
@Override
public double mean()
{
return trials*p;
}
@Override
public double median()
{
if(Math.abs(p-0.5) < 1e-3)//special case p = 1/2, trials/2 is the unique median for trials % 2 == 1, and is a valid median if trials % 2 == 0
return trials/2;
if(p <= 1 - Math.log(2) || p >= Math.log(2))
return Math.round(trials*p);//exact unique median
return invCdf(0.5);
}
@Override
public double mode()
{
if(p == 1)
return trials;
else
return Math.floor((trials+1)*p);
}
@Override
public double variance()
{
return trials*p*(1-p);
}
@Override
public double skewness()
{
return (1-2*p)/standardDeviation();
}
@Override
public double min()
{
return 0;
}
@Override
public double max()
{
return trials;
}
@Override
public Binomial clone()
{
return new Binomial(trials, p);
}
}
Ads help maintain this website.