Using data mining for dynamic level design in games

"Fun" is the most important determinant of whether a game will be successful. Fun can come from challenges and goals, such as victory in a scenario, the accumulation of money, or the right to move to the next level. A game that provides a satisfying level of challenge is said to be balanced. Some researchers use artificial intelligence (AI) on the dynamic game balancing. They use reinforcement learning and focuses on the non-player characters. However, this is not suitable for all game genres such as a game requiring dynamic terrains. We propose to adjust the difficulty of a game level by mining and applying data about the sequential patterns of past player behavior. We compare the performance of the proposed approach on a maze game against approaches using other types of game AI. Positive feedback and these comparisons show that the proposed approach makes the game both more interesting and more balanced.