Calculating P(t_k | !c) when there more than two classes?

[mbaroody]

For the purposes of this project, classification is easy. When we are calculating the ratio score for a 'spam' document, for example, 'not spam' is simply 'ham.' But how do we calculate the conditional probabilities when there are multiple labels that are 'not spam?'

[karthikbmk]

sum over all other classes:

P(t_k|Ck) / P(t_k|C1) + P(t_k|C2).....P(t_k|Ck-1)+P(t_k|Ck+1)+...P(t_k|Cn)

[aronwc]

Correct, but you just need the denominator, i.e.:
Sum( P(t_k|c_j) ) for all c_j != c

On Mar 28, 2016, at 5:52 PM, Karthik [email protected] wrote:

sum over all other classes.

P(t_k|Ck) / P(t_k|C1) + P(t_k|C2).....P(t_k|Ck-1)+P(t_k|Ck+1)+...P(t_k|Cn)

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub