finding specific conditional entropy (for calculating informationgain)

[ssjalakazam]

I am trying to learn how to calculate information gain, and have hit a brick wall. Gain(Y,X) = entropy(Y) - entropy(Y|X)

The first term, entropy, is easy.

But, the entropy(Y|X) is the problem...


so entropy(Y|X) = SUM (prob[x] * entropy(Y|X = x)), over all values of Y.

But what is entropy(Y|X=x)? How do you find it? I have seen nothing online that explains this, and it seems crucial to calculating information gain.