Games Recommender

Project overview

This is the final project of CS114 - Machine Learning @ UIT

In this project, we created a collaborative filtering model based on the matrix factorization model, and a content-based filtering recommender based on the softmax model using the data we collected from Steam. This project is mostly based on this course by Google.

Collect your own data

Create a virtual environment

You should run the scripts in a Python virtual environment. Create venv/ in the project root using:

python -m venv venv

Activate the virtual environment, then run pip install -r requirements.txt to install the dependencies.

Run the scripts

Before doing anything, you should read the README in common/.

First, you will need to scrape the app_ids from Steam:

python -m data.scraper

This script will create a text file called games.txt (specified in common/config.py) containing the scraped app_ids.

Then, collect the games data using:

python -m data.collector collect_games_data

And finally, collect the user reviews data with:

python -m data.collector collect_users_games_data

Sorry for the weird name, it will be changed in the future.

After running the scripts, you should see games.txt, games.csv, and user_game.csv located in data/collected/.

Data

The collected data can be found here.

NOTE: We only collect publicly available data from Steam. Our collected data does not contain any data associated with Steam users, except their steam_id.

Our teams

ID	Name	Github
21520411	Mai Anh Quân	https://github.com/maqnitude
21520456	Trần Xuân Thành	https://github.com/LukasAbraham
21520531	Nguyễn Hà Anh Vũ	https://github.com/AnhVu32