Dota 2 Esports analytics web app. Build with Streamlit, deployed with Heroku
Check it online www.dota4cast.ml
Download:
git clone https://github.com/tarasevich-dmitry/Dota4cast.git
Ensure you have requirements. The project requires a handful of python 3 packages, you can install it with PiP
cd Dota4cast/app/
pip install -r requirements.txt
Then to run app locally:
streamlit run dota4cast.py
Build on results of professional teams performance with TrueSkill algorithm
Pick winrate ML prediction is in dota-2-winner-prediction-on-picks-SGDClassifier.ipynb but it is not ready for production yet
We construct a word cloud to determine the frequency and flavour of the most common words. We get data from in-game all-chat.
Run it online in Google Colab
In tournaments_mapdota_map_parser.py is a parser that takes data from Liquipedia using Beautifulsoup
and then geopy to translate locations to latitude & longitude, final map drawn in streamlit.
Data stored in app and data forlders, however some sorces are quite big and its not possible to load them on github. Also we dont have automated scrappers for all data sources, if you want to help feel free to contribute and improve it!
Initially data taken using OpenDota tools and API's
Aslo check out our Twitter