Riot API Challenge
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
site
.gitattributes
.gitignore
README.md
assigned_normal.zip
assigned_urf.zip
champion_stats.json
champions.json
champions_aggregated_normal.csv
champions_aggregated_urf.csv
champions_assigned.csv
ci_aggregate_normal.js
ci_aggregate_urf.js
ci_assigned.js
ci_download_normal.js
ci_process_normal.js
ci_process_urf.js
ci_rows_normal.js
ci_rows_urf.js
itemfilter.js
itemnames.txt
items.json
items_adjusted.json
items_filtered.json
overall_stats.json
riot.txt
rows_normal.zip
rows_urf.zip

README.md

URF Play Styles - Riot Games API Challenge entry

by Wong Man Hang (NA: microtony, Taiwan: 微Tony)

This project aims to understand how players changed their play style in the URF game mode and the effects. k-means clustering was used to obtain 5 item build centroids (Marksman, Mage, Fighter, Support and Tank). After that, items builds from matches are categorized into one of the 5 centroids. Results are finally aggregated and presented in the mini AngularJS application.

Note: Marksman is not restricted to ranged champions.

Live URL

http://urf.microtony.com

More details of the project can be found in the "About" tab.

Interesting Findings

  1. In URF mode, Ashe was played as a Mage 19% of the time. The win rate is 42.62%.
  2. Building more damage on Poppy gives higher win rate. (Marksman: 61.93%, Tank: 51.97%)
  3. Despite a dramatic shift to Mage builds, for some champions the win rates of such builds did not increase. (Blitzcrank: 51.93% to 49.02%, Malphite: 51.12% to 50.61%)
  4. Thresh has the lowest win rate (about 40%) for all builds.

Technologies used

  • Microsoft Azure ML
  • AngularJS
  • Google Charts
  • Node.JS for downloading data from Riot Games API