package jsat.distributions;
import jsat.linear.Vec;
import static java.lang.Math.*;
import java.util.Random;
import static jsat.math.SpecialMath.*;
/**
*
* @author Edward Raff
*/
public class ChiSquared extends ContinuousDistribution
{
private static final long serialVersionUID = 2446232102260721666L;
double df;//Degrees of freedom
public ChiSquared(double df) {
this.df = df;
}
@Override
public double pdf(double x)
{
if(x <= 0)
return 0;
/*
* df -x
* -- - 1 --
* 2 2
* x e
* -------------
* df
* --
* 2 /df\
* 2 Gamma|--|
* \ 2/
*/
return exp((df/2-1)*log(x)-x/2- (df/2*log(2)+lnGamma(df/2)) );
}
@Override
public double cdf(double x)
{
if(x <= 0)
return 0;
if(df == 2)//special case with a closed form that is more accurate to compute, we include it b/c df = 2 is not uncomon
return 1-exp(-x/2);
return gammaP(df/2, x/2);
}
@Override
public double invCdf(double p)
{
if(df == 2)//special case with a closed form that is more accurate to compute, we include it b/c df = 2 is not uncomon
return 2*abs(log(1-p));
return 2* invGammaP(p, df/2);
}
@Override
public double min()
{
return 0;
}
@Override
public double max()
{
return Double.POSITIVE_INFINITY;
}
@Override
public String getDistributionName()
{
return "Chi^2";
}
@Override
public String[] getVariables()
{
return new String[] {"df"};
}
@Override
public double[] getCurrentVariableValues()
{
return new double[] {df};
}
@Override
public void setVariable(String var, double value)
{
if(var.equals("df"))
df = value;
}
@Override
public ContinuousDistribution clone()
{
return new ChiSquared(df);
}
@Override
public void setUsingData(Vec data)
{
df = ceil(data.variance()/2);
}
@Override
public double mean()
{
return df;
}
@Override
public double median()
{
//2*InvGammaP(df/2,1/2)
return invGammaP(0.5, df/2)*2;
}
@Override
public double mode()
{
return Math.max(df-2, 0.0);
}
@Override
public double variance()
{
return 2 * df;
}
@Override
public double skewness()
{
return sqrt(8/df);
}
@Override
public double[] sample(int numSamples, Random rand)
{
if(df == 2)
return super.sample(numSamples, rand);
//else, lets do different
double[] sample = new Gamma(df/2, 2).sample(numSamples, rand);
return sample;
}
@Override
public int hashCode() {
final int prime = 31;
int result = 1;
long temp;
temp = Double.doubleToLongBits(df);
result = prime * result + (int) (temp ^ (temp >>> 32));
return result;
}
@Override
public boolean equals(Object obj) {
if (this == obj) {
return true;
}
if (obj == null) {
return false;
}
if (getClass() != obj.getClass()) {
return false;
}
ChiSquared other = (ChiSquared) obj;
if (Double.doubleToLongBits(df) != Double.doubleToLongBits(other.df)) {
return false;
}
return true;
}
}
Ads help maintain this website.