[Proj6]Question about the Theory section of Naive Bayes

[Lita Cho]

So I missed the lecture on Thursday, but looking at the lecture slides.
I couldn't find anything that would answer this question.

I was just wondering how this formula:
argmax P(y|f1,...,fm) = argmax (P(y) ∏ P(fi|y))/P(f1...fm)
turned into
argmax P(y|f1,...,fm) = argmax (P(y) ∏ P(fi|y))?

Where did the P(f1...fm) go? I am sure I am missing some rule or
something. I apologize for my horrible notation. The equations can be
found in the Theory section of the Naive Bayes in the Project 6 website.

Lita

[Daniel Ritchie]

In the first equation you wrote, the argmax is over a conditional
distribution. In the second, the argmax is over a joint distribution.
The difference between them is that the conditional distribution is
normalized. But since normalization just multiplies every entry of the
joint by the same constant value, the relative order of entries remains
the same. Thus argmaxing over either gives the same result--whatever
entry has the largest value in one will have the largest value in the other.

Post by Lita Cho
So I missed the lecture on Thursday, but looking at the lecture slides.
I couldn't find anything that would answer this question.
argmax P(y|f1,...,fm) = argmax (P(y) ∏ P(fi|y))/P(f1...fm)
turned into
argmax P(y|f1,...,fm) = argmax (P(y) ∏ P(fi|y))?
Where did the P(f1...fm) go? I am sure I am missing some rule or
something. I apologize for my horrible notation. The equations can be
found in the Theory section of the Naive Bayes in the Project 6 website.
Lita