This is a Part II project I have done at uni for my disseration.
It aims to predict what results different players are gonna get in a match based on their history.
More information about the whole process in dissertation itself: Dissertation/Dissertation.pdf.
In particular, chapter 3 explains the reasoning behind the model.
Code for data collection in is 'match_data_collection'
Code for the prototype is in 'data_analysis'
Code for final system is in 'prediction_model'
I did not include match data as it is about 300MB, if you decide to test the system, put raw json files in 'all_matches' and update 'match_stats.json'. Some of the data I used can be found here.