# Possible dictating mistake in logistic regression video

Do correct me if I am wrong, but I think there might be a slight mistake in the video Cost function for logistic regression in week 3

At time 9:19, Andrew says " In fact, if f of x approaches 0, the loss here actually goes really large and in fact approaches infinity."

But this is not true because in the case of actual y = 0, then if f(x) approaches 0, loss also approaches 0 as it is also written at 8:15.

Though perhaps he was taking in some other context. Not sure that’s why I came here.

You are saying that “If y = 0, then as f(x) → 0, then the loss → 0”. This is not wrong.

SInce you were also wondering if Andrew might be discussing for other case (which can only be y = 1), do you think that the narration between 9:09 - 9:13 is unclear for which case is being discussed?

Cheers,
Raymond

Hello @rmwkwok

Thank you for your reply. Let me clarify further. I understand completely that in the case y = 0, then as f(x) → 0, then the loss → 0.

But at time 9:19, Andrew says “In fact, if f of x approaches 0, the loss here actually goes really large and in fact approaches infinity.”

These statements are contradictory hence the confusion.

You said this was talked at 8:15.

You said this was talked at 9:19.

So, (Q1) between 8:15 and 9:19, did Andrew change from talking about y=0 to talking about y=1? (Q2) If you listen to 9:09-9:13, can we be 100% sure that Andrew did NOT switch to y=1? (Q3) If he was indeed talking about y=1, would it be still contradictory?

To me, the answers to Q1 and Q2 are not certain and we need the course team to clarify, but for Q3, I think we know the answer is NO, right?

Cheers,
Raymond

@rmwkwok
Yup! That’s exactly what I am talking about. I too believe that Q1 and Q2 need to be clarified and the answer to Q3 is indeed NO.

Great, @i200660_Mirza_Ubaidu! Now we are on the same page - your understanding is correct, but the video sounds confusing. In fact, I had shared this with the course team, and then we have to wait for their response!

Cheers,
Raymond