Abstract
Balancing video game difficulty is a constant struggle to keeping the game challenging to maintain player interest but not overly challenging as to frustrate the player. While the player progresses through the game, their perception of too hard and too easy changes. Games often use a static difficulty curve to increase the difficulty as the player progresses through the game. However, dynamically adjusting the game difficulty creates a better fitting difficulty balancing, leading to a better player experience. To create such an experience, this work focuses on machine learning techniques \textbf{decision trees} and \textbf{Fuzzy Inference Systems} for the use of Dynamic Difficulty Adjustment. A decision tree learns how to evaluate a player's performance based on in game statistics and outputs a ranking. This ranking is then used by the Fuzzy Inference System, along with the player's current health and damage output, to decide if the difficulty is increased, decreased, or left the same and to what degree it is changed. Therefore, a combination of decision trees and Fuzzy Inference Systems lead to a Dynamic Difficulty Adjustment.
Advisor
Byrnes, Denise
Department
Computer Science
Recommended Citation
Bonadio, Michael Vincent, "Dynamically Adjusting Video Game Difficulty Using Machine Learning Techniques" (2016). Senior Independent Study Theses. Paper 7230.
https://openworks.wooster.edu/independentstudy/7230
Publication Date
2016
Degree Granted
Bachelor of Arts
Document Type
Senior Independent Study Thesis
© Copyright 2016 Michael Vincent Bonadio