League of Legends match history analysis completed as part of social computing course project. Some of this code was written hastily, so there are some horribly copy/pasted areas (entire copies of graph.py for example).
There is also a bug with pullData.coffee (see below for more info) that I worked around via manual intervention for the sake of time/delivery.
For pulling data from league API - must have mongodb running.
(For graphing) Create an account with Plotly - https://plot.ly/python/getting-started/ and setup credentials, api key, etc.
pip install -r pyScripts/requirements.txt
To pull API data
npm install
coffee pullData.coffee
Matches are written to mongo, the database will be named 'soc-lol-analysis' database with 'players' collection. Note: There is a bug I didn't take the time to fix - the program hangs when trying to pull multiple players data, it can only pull one player at a time. So, I pulled for one player, commented out that player, then ran for the next, and so on (for 10 players). If extending this further to pull 100s or 1000s of players' match data, obviously this would need to be fixed.
players/<name>.json files are just sample data pulled from mongo after the API pull (using some simple bash)
Before running scrub.py, I manually (bash) removed the ObjectID field from the resulting files so that they are truly JSON instead of BSON. This was me being lazy and not wanting to parse BSON.
To run scrub.py
scrub.py <file 1> ..<file n>
<file 1>.scrubbed ... <file n>.scrubbed will be created.
players/<name>.json.scrubbed files contain one line per game in the format of
<name> <win> <timeSinceWin> <timeSinceLoss>
To graph
pyScripts/graph<version>.py players/<name>.json.scrubbed
For the most interesting graph(imo), use graph_diff_30.py
The resulting graph(s) should pop up in a new tab(s) of your default browser if you have everything configured properly.
A PDF of my analysis/writeup is included here