smn-ailab / pycmf Goto Github PK

This tanh implementation of sigmoid is two times faster than scipy.special.expit.

like we have 1000 users, 2000 movies and each movie have some tags, total tags is 200.

and we have user rating, from 1 to 5

the train_x maybe users+movies+tags, like nx3200 matrix, and train_y maybe nx1

so we got U, V, Z from cmf model

and we can prediction score from V@Z_T

am I right?

The cython extension needs to be built upon install. Currently running

$ python setup.py install

completely ignores the cython extensions.

setup.py seems to be installing the package, but python fails to discover it...