Conditional Entropy

Calculates the conditional entropy of a discrete variable, given other discrete variables.

Description

ConditionalEntropy: Calculates conditional entropy (in bits) of Y, given X

  H = ConditionalEntropy(Y,X)
 
  H = calculated entropy of Y, given X (in bits)
  Y = dependent variable (column vector)
  X = independent variable(s)
  
  Note 1: Each distinct value is considered a unique symbol.

  Note 2: requires 'Entropy' and 'MutualInformation' functions
 
  Example: 'X' (1 bit of entropy), 'Y' (2 bits of entropy)
    Given 'X', 'Y's conditional entropy is 1 bit.
 
  Note: Estimated entropy values are slightly less than true, due to finite sample size.
 
  Y = ceil(4 * rand(1e3,1)); X = double(Y <= 2);
  ConditionalEntropy(Y,X)

• MATLAB release: MATLAB 6.0 (R12)
• Other requirements: Requires function 'MutualInformation'.