https://github.com/Wadenschwinger/LoLFirsts
1st place @ Predictive Analytics Hackathon, Hannover, 11.11.2017
https://github.com/Wadenschwinger/LoLFirsts
analytics classification decision learning machine predictive tree
Last synced: 4 months ago
JSON representation
1st place @ Predictive Analytics Hackathon, Hannover, 11.11.2017
- Host: GitHub
- URL: https://github.com/Wadenschwinger/LoLFirsts
- Owner: SanRehmo
- Created: 2017-11-11T16:48:31.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2017-11-13T14:36:37.000Z (over 8 years ago)
- Last Synced: 2024-12-28T02:35:20.296Z (about 1 year ago)
- Topics: analytics, classification, decision, learning, machine, predictive, tree
- Language: Python
- Homepage:
- Size: 3.65 MB
- Stars: 0
- Watchers: 4
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# LoLFirsts
This is a classifier we wrote for the Predicitve Data Analysis Hackathon in Hannover, 11.11.2017. The goal was to extract something useful from a League of Legends dataset.
We trained a Binary Decision Tree on the big "firsts" of each match:
* firstBlood
* firstTower
* firstInhibitor
* firstBaron
* firstDragon
* firstRiftHerald
Using these, we predict if a team will win the match. The classifier gets to around 87% accuracy on our test split. If we take out the Inhibitor, which appears late into the game, we still get over 70%. Still enough to justify rage quitting, I'd say.