What is the formula for finding mutual information (MI) feature to feature? Is this formula correct $$ \mathrm{MI}=p(t_1,t_2)\log_2\left(\frac{p(t_1,t_2)}{p(t_1)p(t_2)}\right) + (1-p(t_1,t_2)) \log_2\left(\frac{1-p(t_1,t_2)}{(1-p(t_1))p(t_2)}\right)?$$ i am doing a project in sentiment analysis.here pt1t2 is the probability of cooccurence of t1t2 together.so is in the part of finding probabilty of not occuring together,is in the denominator (1-pt1)*(1-pt2) or as given in the equation .
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
This has already been asked and here is a very good answer with a code:
