Análise das características de jogabilidade no PUBG usando árvore de decisão

Abstract

The present work developed with the database available by Kaggle, aims to identify, using the Decision Tree (AD) model, a compiled strategies used by players who have significant wins insolomodeoftheonlinegamePlayerunknown’sBattlegrounds(PUBG).InPUBG,solomode is supported by 100 players in each match. The game consists of an island delimited with areas of automatic death by the game, as well as a circular security area that narrows during the game, as well as other players that can exterminate the opponent. Each player will have access to attack and protective instruments. The goal is to be the last one alive at the end of the match regardless of the strategies used. To achieve our purpose, a literary base on the subject was first structured, then an exploratory analysis of the data to verify possible outliers and the distribution of variables and finally an analysis of the Decision Tree model. From the helper software, Knime was chosen for exploratory data analysis and for the execution, validation and interpretation of the model, the software Knime and R. The base was redistributed in 70 % and 30 %, to train the model and validate it. respectively. Using the AD technique, it was possible to understand the profile of the players who tend to have a low chance of survival, as well as the fittest, with accuracy of 0.84 for the model obtained.