Image by d3images on Freepik
GameGuard is a machine learning solution designed to prevent video game addiction linked to microtransactions. It tracks and mitigates addictive behavior in real-time by monitoring and analyzing user purchase patterns for loot boxes, a kind of microtransaction in which the item the player obtain is randomized.
The use of microtransactions, especially in loot boxes, can lead to addictive behavior and excessive spending among players. GameGuard was created to address the problematic aspects of microtransactions that negatively affect the game developer industry and to provide a solution that considers the mental health issues associated with video games.
GameGuard uses open data about loot box purchases for the game Counter Strike in the Chinese market, which includes information on the timestamp, username, and outcome of each purchase. The value for each lootbox and its outcome has been scrapped from an online market where people buy and sell these lootboxes.
It also uses a labeled dataset on online gambling to train a machine learning classifier that identifies users who are at risk of addiction in the lootbox dataset.
The solution is accessible through an online dashboard, which provides an overview of the lootbox purchase patterns and shows the users with the higher risk score. On each user's profile page, the evolution of the risk score through time is shown, together with daily statistics of their purchase history, with the possibility of inspecting the raw data. All those features are designed to allow regulatory agencies to take necessary actions to prevent addiction.
Alright, let's see GameGuard under the hood.
For an online demo, just head to http://gameguard.marcpalaus.com
Otherwise, follow the Installation instructions below to run GameGuard in your own machine.
- ⭐ Star this repository if you like this project!
- 🧪 Testing: Try it! Follow the installation instructions and try it for yourself.
- 🐞 Issue reporting: Found a bug? Help us improve by reporting it!
- 💡 Request features: Do you miss something important? Submit a feature request!
To avoid conflicts with the versions of the dependencies, it is recommended that GameGuard is installed in its separate environment
git clone https://github.com/wonx/lootbox_addiction
cd lootbox_addiction
pip install --user virtualenv
python3 -m venv .gameguardenv
source .gameguardenv/bin/activate
pip install -r requirements.txt
e.g. apt install screen for Debian-based systems.
screen -dmS "lootbox_launcher" python launch.py
It will create two background screen sessions, lootbox_launcher and lootbox_flask (you can see and enter them by using the screen -r parameter). The first one, lootbox_launcher is the scheduler for the background scripts that will collect and process the data, and the second one, lootbox_flask is the process for the webapp.
You can access the dashboard by browsing to http://localhost:9018
Data will take some time to populate, so not everything will be available in the dashboard immediately. The background ingestion script by default will gather data every 10 minutes, computer the lootbox values every 30 minutes, and compute the risk score for the users on sundays at midnight (Shanghai timezone).
If the webapp refuses to start due to the dataframes not being yet ready, you manually start it later on by running: screen -dmS "lootbox_flask" python3 flask_app/app.py