Predict Players Engagement Level using KNN Method

Abstract

According to the video game industry development, players behavior analysis in game corporation becomes more essential than before, which not only helps game designer know the market needs, but helps company making future decision of the game as well. Therefore, in this experiment, the author built a model for machine learning about players behavior, a dataset found on Kaggle, which contains 40, 034 players with 13 features of each player such as Player ID, Age, Gender, Location, Game Genre, Play Time Hours, In Game Purchases, Game Difficulty, Sessions Per Week, Avg Session Duration Minutes, Player Level, Achievements Unlocked and Engagement Level. This experiment will use K-Neighbor (KNN) algorithm to make a model with random forest and some evaluation metrics to evaluate the model. The results of this research are not only valuable for predicting player engagement but can also provide actionable insights for game companies, such as optimizing monetization strategies, improving game design, and enhancing personalized user experiences.