Date of Degree
Data Analysis & Visualization
Matthew K. Gold
eSport, Tableau, Pick or Ban, Hero recommendation
This research explores how to choose heroes for the popular online eSport game Dota 2. Hero selection, also known as drafting, is so crucial to the game that carefully designed hero choices can “implicitly give a team a large advantage before the match even begins” (Conley and Perry 2013).
Different combinations of heroes in a team will have different interactions, ultimately yielding a variety of match outcomes. By combining data visualization and neural network machine learning, this project seeks to help players choose heroes to maximize a team’s likelihood of victory. I will start by introducing some game-play terms. I then visualize key statistics of heroes and interpret how to use these visualizations to choose heroes. Next, I use a basic neural network machine learning model to predict a team’s match outcome (win/lose) based on their hero choices. Future research should use a larger dataset to maximize the model’s accuracy and incorporate each hero’s performance (GPM, XPM, DPM, BDPM, and win rate) into the machine learning model.
Gong, Zhan, "Dota 2 Hero Selection Analysis" (2021). CUNY Academic Works.