question

[rgap]

hi.

in this question

http://www.youtube.com/watch?v=DjvGl1qRVdE&feature=related

why its not P(SPAM) = (9+1)/(24+2) ???????????????

[benoît person]

Hi,

Because, by "spam" we are talking about "message" not "word". So, when you count, you have 3 spams messages for a total of 8 messages, not 9 spam-words for a total of 24 words.

--
You received this message because you are subscribed to the Google Groups "Stanford AI Class" group.
To post to this group, send email to stanford...@googlegroups.com.
To unsubscribe from this group, send email to stanford-ai-cl...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/stanford-ai-class?hl=en.

--
Benoit Person

Ensimag - 1ère année

[benoît person]

Even if we were counting "words", your formula would be wrong, because it would be : "9 + 1 / 24 + 1*size_of_the_vocabulary". And, "size_of_the_vocabulary" would be 13 if you were considering "word". Here we use "2" because a message is a spam or not.

2011/10/31 benoît person <benoit...@gmail.com>

[Anders Alm]

Because this question is focusing on any random message from the list as a whole, not the words.

In this case I see P(SPAM) as; if you pick any defined message from the list (without knowing what message you pick), what is the probability of it being spam.