I've used Python quite extensively, but after taking three classes that used "R", my impression is that you ignore R at your peril. More emphasis on RPy may be the answer (or some way to incorporate the domain expertise of the legions of R package developers). From this talk, my impression is that Python also needs "PyDoop".
I agree that numpy needs to stream arrays and matrices better. The way I understand it, this is not possible at the moment. The __array__() (or whatever it is called) method/property of an object, cannot give other methods access to an array without loading all of it into memory. for me this also meant that as a matrix is essentially a symmetric array of arrays, all of it had to be loaded. You only get the syntactic convenience, unless I am wrong about something!
I've used Python quite extensively, but after taking three classes that used "R", my impression is that you ignore R at your peril. More emphasis on RPy may be the answer (or some way to incorporate the domain expertise of the legions of R package developers). From this talk, my impression is that Python also needs "PyDoop".
I agree that numpy needs to stream arrays and matrices better. The way I understand it, this is not possible at the moment. The __array__() (or whatever it is called) method/property of an object, cannot give other methods access to an array without loading all of it into memory. for me this also meant that as a matrix is essentially a symmetric array of arrays, all of it had to be loaded. You only get the syntactic convenience, unless I am wrong about something!
Thanks. Conceptual clarity.
thank you.
it's not talking about big data at all, let alone data analysis on big data using python. what a waste of my time.
lol wang