Outcome Prediction and Feature Importance Analysis of Immortal-rank Dota 2 Games

Date
2021-01-01
Authors
Boyle, Tanner
Major Professor
Zhengyuan Zhu
Advisor
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Authors
Research Projects
Organizational Units
Statistics
Organizational Unit
Journal Issue
Series
Department
Statistics
Abstract

Defense of the Ancients 2, better known as Dota 2 or just Dota, is a popular eSport video game known for both its learning curve and high skill ceiling. Its professional players are the highest paid out of any video game and prize winnings from single tournaments in Dota 2 are equivalent to years of competition in other eSports. Paired with this, interest from casual players in high-level strategies has been rapidly increasing – as seen by the growing number of educational media sources (Twitch live streams and YouTube channels) and professional players offering paid coaching sessions. This study utilizes a uniquely constructed dataset and feature space to look at the first 20 minutes of the highest skill games in the world and predict the eventual winner. Additionally, we look to identify the features that contribute most to a strong prediction in hopes of advancing the public knowledge of statistically optimal strategies.

Comments
Description
Keywords
Citation
DOI
Source