AI-based Player Behavior Prediction and Game Experience Optimization: A Study on Classification, Regression, and Deep Reinforcement Learning Models Combining Behavioral and Social Data

Based on behavioural and social data, this research focuses on the application of AI models for predicting player behaviour and optimizing game experience. This research uses classification, regression, and deep reinforcement learning (DRL) models to anticipate the player behaviour, aid in-game decision making, and customize game environments. This coupling of behavioural data like session duration, activity frequency, decision history and social data like player interactions, group dynamics, communication logs is essential for accurate predictions. This research tests various ML models such as decision trees, SVM, random forests, and deep reinforcement learning to give an in-depth understanding on how these models support the experience of the players. In the part of experimentation, this research has demonstrated that, while classification and regression models are excellent for predicting discrete player behavior and continuous game variables, DRL models are better for predicting real-time game optimization. This insight also points to the need to integrate behavioural and social data to develop dynamic, interactive and personalised gameplay. The research provides some insights for future AI-based games applications especially for creating mechanisms that continually change in response to the user.