Skip to content

Using Embeddings and DNNs to predict outcomes of Dota 2 matches

Notifications You must be signed in to change notification settings

talhasaruhan/dota2

Repository files navigation

Machine Learning Applications on Dota 2

The suggestion model is currently being migrated

Models:

  • train_hero_embeddings.py : model for representing Dota 2 heroes in a latent space, includes logger and saver utilities.
  • train_predictive_model.py : model for predicting Dota 2 match outcomes using the pre-trained hero embeddings. Includes utilities.

Other Utilities:

  • OpenDotaAPI.py : Partial Python wrapper for OpenDota API

Data:

  • pro_dump.p : professional match data until Aug. 17. In the format: [match_id, start_time, radiant_win, [heroes]]
  • final_embeddings : trained embeddings on pro_dump.p

About

Using Embeddings and DNNs to predict outcomes of Dota 2 matches

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Languages