I understand why in the videos, Professor Ng teaches the logistic regression as w.T, because the convention is that all stand alone vectors are column vectors (thanks to Paul for helping me understand that). In the coding assignment however for week 2, could I just initialize my w vector to be (1,nx) and skip the transposition step? Or would that mess things up?
You’re welcome to try it and see what happens. You’re right at logical level that you could do that and it should work in terms of the final output. But once you look at how the assignment is structured, you’ll find that they test all the functions separately, e.g. the “initialize” function and the “propagate” function. So they enforce Prof Ng’s way of doing things.
Sorry, but Prof Ng is the boss here. When you’re the Professor teaching the class, you can do things your way, but for now you have to do things Prof Ng’s way. 
Haha fair enough. Thanks for your help!