Riot API Challenge 2016 (spring)
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
codes
images
output
Documentation.md
HowTo.md
README.md
riot.txt

README.md

MasteryMakeMaps

Riot API Challenge 2016 (spring)

by Lyunee

  • This is my submission to the RIOT API challenge 2016, following the API terms and condition
  • The website Mastery Makes Maps presents this entry's results.
  • The documentation contains the detailed ideas and processes.
  • The HowTo document presents the steps to follow in order to reproduce the results.
  • The codes are in the "code" folder
  • The folder "output" contains the files generated through the analysis of the data, as well as the files used for the presentation of the results in the website.
  • The folder "images" contains a few additional images used in the documentation.
  • The programs were written in python, and run using python 2.7.10

Those programs were written to run as a python backend of the website, but the backend-to-frontend was not implemented. The backend processes were thus run locally and the results presented in the website Mastery Makes Maps are static.


This entry uses the champion mastery points to create a graph (or map) of the champions. This graph is based on the "similarity" of the champions from the players' point-of-view.

It can be analyzed to extract groups of similar champions. This is achieved using community detection. The groups correspond to in-game positions and are further analysed : is there sub-groups? (yes except for the ADC group), can we find bridges between the groups? (yes)

The graph can also be exploited to suggest ways to expand a player's champion pool. In particular, using closest neighbors, we can answer the question "What would be easy to learn next after mastering a given champion?"