Map classification ported from a jupyter notebook implementation by Yoeri Poels.
Gives maps a similarity value by comparing top 50 times on every map and class. Recommends maps to players based on their existing top times. Only works for players who have top times on Tempus.
Current implementation is hosted at laurirasanen.github.io/tempus_recommend.
You can also navigate directly to laurirasanen.github.io/tempus_recommend/recommend/<player_id>/<class_id>, where class id is 3 for soldier or 4 for demoman.
You can find your player id over at tempus.xyz as a part of the url after searching for a player.
Update db and train model:
python update.py
python train.py
- for flask hosting:
python app.py - for static sites:
python freeze.py