This is a python code for probabilistic matrix factorization using hand-written SGD update rules in recommendation.
All files are organized and they are easy to be understood
You can use movielen-1m for testing this code. Please note the data path in this code are all relative path.
The files are following:
===>1. 0.data_process-1.py
Generate data for pmf_main.py file
===>2. pmf_main.py
The main file of pmf algorithm, define some hyper-parameters.
===>3 pmf_model.py
This file contains the main pmf model definition.
===>4 evaluations.py
This file defines the evaluation metric way for this algorithm.(RMSE in this file)
Runing Note:
0.data_process-1.py ---> pmf_main.py
paper reference:
Probabilistic Matrix Factorization
http://papers.nips.cc/paper/3208-probabilistic-matrix-factorization.pdf