Automated Rating of Level Difficulty for Puzzle Games

Abstract

In this thesis, a method for automatically rating the difficulty of puzzle game levels is introduced. Our method takes multiple aspects of the levels of these games, such as level size, and combines these into a difficulty equation. The method can simply be adapted to most puzzle games. We test this method on three different puzzle games: Flow, Lazors and Move. We conducted a user study to find out how people rate the difficulty of a set of levels. We then use this data to train the difficulty equation to match the user-provided ratings. Our experiments show that the difficulty equation is capable of rating levels with an average error of approximately one point in Lazors and Move, and less than half a point in Flow.