GoTo School @ Summer 2016 baseline

(School announcement on habrahabr.ru (in Russian), with additional school description and extended explanation of methods used.)

This repository features two baseline versions (both based on collaborative filtration): a simpler one (baseline-simple.ipynb) based on cosine similarity between film audiences and a more complex one (baseline-intermediate.ipynb) using TSVD user-item matrix decomposition. There's also a notebook with simple examples on data manipulation (opening, mostly) with an intuitive name of data_manipulation.ipynb.

The metric used is mAP@10 (NB: be sure to check the notes in the notebooks before comparing scores!).

Using

git clone $project
cd $project

virtualenv devenv
source devenv/bin/activate

make deps

make download_data extract_data

jupyter notebook

Licensed under Apache 2 and MIT.