I’m currently doing another ML course that is using matlab/octave, so that’s also hot on my mind. I find matlab more pleasant to use for simple matrix manipulations, so I mentally code in matlab and then translate to numpy.
I found these pages helpful for translating.
http://mathesaurus.sourceforge.net/matlab-numpy.html
https://docs.scipy.org/doc/numpy-dev/user/numpy-for-matlab-users.html
The most recent head-banging problem that I hit was that numpy distinguishes between arrays and 1-D matrices; transposing an array doesn’t do anything! 