This research aims to create a new predictive model for estimating player engagement levels and assess the effectiveness of the Random Forest Classifier applied in the study. To achieve this, the research uses the “Predict Online Gaming Behavior Dataset” provided by Rabie El Kharoua on the Kaggle platform ( www.kaggle.com ). This dataset offers detailed metrics and demographic information regarding player behavior in an online gaming context. The experimental findings show that, among all variables, TotalPlayTimePerWeek, AvgSessionDurationMinutes, and PlayerLevel were identified as the most significant factors influencing engagement through feature importance analysis. Additionally, the model exhibited excellent performance with an accuracy of 90.78%, a precision of 90.80%, a recall of 90.78%, and an F1 score of 90.74%. This research contributes both theoretically and practically. First, it expands the understanding of player engagement by empirically analyzing the combined effects of different variables on engagement modeling. Second, the methodology outlined in the paper provides a framework for those facing similar challenges. Lastly, the model proposed in this study allows game developers to forecast player engagement levels ahead of time during game operations, facilitating the development of marketing strategies, improved player experiences, and adjustments to game mechanics to enhance player engagement.