/DotaDraftAP

Machine learning application for drafting All Pick matches in Dota 2.

Primary LanguagePython

DotaDraftAP

Machine learning application for drafting All Pick matches in Dota 2.

Getting started

You can configure DotaDraft by editing the parameters in config.py.

Getting training data

In order to fetch new match data for training, STEAM_API_KEY must be set to your Steam Web API key in config.py.

To make sure the hero information is up-to-date, run python heroes.py.

Run python matches.py and new matches will begin to be fetched (this is very slow) and output to file as JSON.

The data is formatted as

{
    "data_size": 1,
    "matches": [
        {
            "match_id": 0,
            "match_seq_num": 0,
            "radiant_win": true,
            "game_mode": 0,
            "lobby_type": 0,
            "picks_radiant": [1, 2, 3, 4, 5],
            "picks_dire": [6, 7, 8, 9, 10]
        }
    ]
}

Run python process.py to generate a JSON file containing data converted for use as training data.