DotA 2 bots win prediction using naive bayes based on adaboost algorithm
Pulung Nurtantio Andono, Nanang Budi Kurniawan, Catur Supriyanto · Proceedings of the 3rd International Conference on Communication and Information Processing · 2017
DotA 2 is a multiplayer game that is widely played today. In DotA 2, the players are divided into two teams, i.e., radiant and dire to against each other. Each team consists of five heroes. One hero is played by human and four heroes are controlled by artificial intelligence (AI). In our prediction, we only collect the statistical data of AI heroes. Each team has the main headquarters, which needs to be protected; the headquarters is a called ancient. When the ancient of a team is destroyed, then the game is over. The features of prediction are collected from the statistical value in each bot. Therefore, the player knows which team (radiant or dire) is going to be the winning team. To predict the winning team, we use Naive Bayes (NB) as a classifier in data mining algorithm. NB is the appropriate algorithm. Since NB works on probability, it can predict the winning team, not only the winning heroes. However, NB has a shortcoming, e.g., imbalance data, this study proposes to implement NB+Adaboost. This study evaluates some approaches of NB, i.e., discretization and Gaussian distribution kernel function. Both are used to treat the numerical attribute. The results of the experiment show that the highest accuracy of the win prediction by using NB+Adaboost with Gaussian distribution kernel achieves 80%.