I totally have no idea Y_prediction. How should I do ?
If there is a video talking about this problem, please tell me.
They explain what to do in the instructions for that section. You compute the output of the model for the input matrix. That gives you values that are the output of sigmoid for each input image, right? So they are between 0 and 1 and look like a probability. So we interpret it as 1 (yes it’s a cat) if the probability is > 0.5 and 0 (no it’s not a cat) if the probability is <= 0.5. They even wrote out most of the code for you in the template: they give you the for loop and a pretty big hint about what goes in the loop in the comment there.
Thank your for reply ! I’m Japanese and sometimes these kinds of technical jargon drive me crazy !
Learning deep learning with English is so hard !
Thank you !
Thanks for your response. Sorry! Sometimes I forget that the issue may be more about natural language than computer language! Everybody understands python, but English is a whole lot more complicated than python. 
But the high level lesson here is that the instructions are usually pretty complete. If you are feeling lost about what to do, my first step is always to read the instructions again. Also sometimes you get kind of stuck by thinking about something in the wrong way and it’s hard to break out of that. Just going for a walk or getting something to eat and then coming back and reading again with “fresh eyes” can really help to see in a new way!