SaltieRL / ReplayModels

:car: Machine Learning models for Rocket League

ReplayModels

Creates models for Rocket League replay analysis.

Implementation

General classes and modules

Data retrieval is done through the DataManager subclasses (e.g. CalculatedLocalDM). These subclasses expose a get_data() method which retrieves a GameData object (with .df and .proto attributes).

General utility functions such as filtering columns of the dataframe are available in data/utils/utils.py.

data/utils/number_check.py checks the number of available replays in calculated.gg's api for a certain query, for a given playlist and min MMR.

value_function

Use batched_value_function.py which uses the refactorised class BatchTrainer. Running it should cache replay dataframes and protos, and plot loss with quicktracer.

data_main.py

Run this script to either download replay files, convert them to CSV, or combine CSVs into a dataset. Doing this relies on the config.ini file in "data/"

Steps to getting a dataframe of replay data:

Set up your config.ini file (what mode and mmr range you want to deal with, path options)

In the command line with the necessary packages installed:

    python data_main.py      (to see what args you want to use)
    
    python data_main.py download [args]
    python data_main.py convert [args]
    python data_main.py dataset [args]

You now have a .h5 file that can be opened by pandas into a dataframe.