Recommender Systems 2016

University project for the course of Recommender Systems 2016 hold by Prof. Paolo Cremonesi at Politecnico di Milano

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes.

Prerequisites

In order to run the project you'll need Python version 3 and above (LINK) You will also need this pip packages:

numpy==1.12.0+mkl
pandas==0.19.2
scipy==0.18.1

Installing

After having installed Python just run

pip install numpy pandas scipy

When everything is set up just run

python tf-idf-ratings.py

You should get something like

Loading data...
Data loaded in 0:00:01.081037
Prepocessing Data...
Ended preprocessing in 0:00:00.117112
Starting recommending!

User 285
[749025, 827566, 1808563, 1763102, 1965165]
User 285 computed in 0:00:04.406224
...

The output will be stored in the file test.csv in the csv format like:

user1,rec1,rec2,rec3,rec4,rec5
user2,rec1,rec2,rec3,rec4,rec5

###Troubleshooting

If you miss some of the file the files in the directory precomputedData you can generate them doing the following:

userRatingSimilarity_IP.npz

Uncomment lines 9-14 in utils\cfutils.py
Comment line 18 in utils\cfutils.py
Run function create_user_rating_matrix_similarity in utils\cfutils.py

user_rating_matrix.npz

Uncomment lines 9-14 in utils\cfutils.py
Comment line 18 in utils\cfutils.py
Run function create_user_rating_matrix in utils\cfutils.py

Branches

The master branch holdes the code that scored most in the competition, every other branch is named by the algorithm used to create the recommendations

Coding style

Whenever possible the code tries to follow recommendations specified by PEP8

###Data and Submissions The data can be downloaded from the platform Kaggle

All the submission done to the platform Kaggle can be found in the folder submission tagged with the date

Authors

Paolo Paterna
Lorenzo Zoia

TopoDiFogna / recsys2016