Difference between revisions of "Dirichlet"

m (adding doc status category)
Line 2: Line 2:
 
[[Category:Doc Status C]] <!-- For Lumina use, do not change -->
 
[[Category:Doc Status C]] <!-- For Lumina use, do not change -->
  
= Dirichlet( alpha,I ) =
+
== Dirichlet(alpha, I) ==
  
A Dirichlet distribution with parameters alpha_i>0.
+
A Dirichlet distribution with parameters «alpha»<sub>i</sub> > 0.
  
Each sample of a Dirichlet distribution produces a random vector whose elements sum to 1.  It is commonly used to represent second order probability information.
+
Each sample of a [[Dirichlet]] distribution produces a random vector whose elements sum to 1.  It is commonly used to represent second order probability information.
  
 
The Dirichlet distribution has a density given by  
 
The Dirichlet distribution has a density given by  
k * [[Product]]( X^(alpha-1), I)
+
:<code>k*Product(X^(alpha - 1), I)</code>
 +
 
 
where k is a normalization factor equal to
 
where k is a normalization factor equal to
[[GammaFn]]( [[Sum]](alpha,I )) / [[Sum]]([[GammaFn]](alpha),I)
+
:<code>GammaFn(Sum(alpha, I))/Sum(GammaFn(alpha), I)</code>
  
The parameters, alpha, can be interpreted as observation counts.  The mean is given by the relative values of alpha (normalized to 1), but the variance narrows as the alphas get larger, just as your confidence in a distribution would narrow as you get more samples.
+
The parameters, alpha, can be interpreted as observation counts.  The mean is given by the relative values of alpha (normalized to 1), but the [[variance]] narrows as the alphas get larger, just as your confidence in a distribution would narrow as you get more samples.
  
The Dirichlet lends itself to easy Bayesian updating.  If you have a prior of alpha0, and you observe N
+
The [[Dirichlet]] lends itself to easy Bayesian updating.  If you have a prior of «alpha0», and you observe ''N''.
  
= Library =
+
== Library ==
 +
Multivariate Distributions.ana
  
Multivariate Distributions.ana
+
==See Also==
 +
* [[GammaFn]]
 +
* [[Multivariate distributions]]
 +
* [[Distribution Densities Library]]

Revision as of 00:47, 28 January 2016


Dirichlet(alpha, I)

A Dirichlet distribution with parameters «alpha»i > 0.

Each sample of a Dirichlet distribution produces a random vector whose elements sum to 1. It is commonly used to represent second order probability information.

The Dirichlet distribution has a density given by

k*Product(X^(alpha - 1), I)

where k is a normalization factor equal to

GammaFn(Sum(alpha, I))/Sum(GammaFn(alpha), I)

The parameters, alpha, can be interpreted as observation counts. The mean is given by the relative values of alpha (normalized to 1), but the variance narrows as the alphas get larger, just as your confidence in a distribution would narrow as you get more samples.

The Dirichlet lends itself to easy Bayesian updating. If you have a prior of «alpha0», and you observe N.

Library

Multivariate Distributions.ana

See Also

Comments


You are not allowed to post comments.