Games and Big Data: A Scalable Multi-Dimensional Churn Prediction Model
Keywords:
social games, churn prediction, ensemble methods, survival analysis, online games, user behavior, big dataAbstract
The introduction of mobile games has resulted in a paradigm shift in the video game business. Game developers now have a wealth of information on their players at their disposal, allowing them to use trustworthy models that can reliably predict player behavior and scale to massive datasets. Churn prediction, a difficulty shared by many industries, is especially important in the mobile gaming business, where user retention is critical for effective game monetization. We offer a method for predicting game churn based on survival ensembles in this article. Our algorithm accurately predicts both the level at which each player will abandon the game and their total playtime up to that point. It is also resistant to varied data distributions and adaptable to a wide range of response variables, while allowing for effective parallelization of the algorithm. As a result, our methodology is well adapted to doing real-time churner assessments, even for games with millions of daily active users.
References
Hicham N, Nassera H, Karim S. Strategic framework for leveraging artificial intelligence in future marketing decision-making. Journal of Intelligent and Management Decision. 2023;2(3):139-50.
Martínez-Carrascal JA, Sancho-Vinuesa T. Impact of Assessment Characteristics on Course Withdrawal: a Survival Analysis Approach.
Prentice RL, Zhao S. Regression models and multivariate life tables. Journal of the American Statistical Association. 2021 Jul 3;116(535):1330-45.
Weiss I, Vilenchik D. Predicting Churn in Online Games by Quantifying Diversity of Engagement. Big Data. 2023 Aug 1;11(4):282-95.
Hothorn T, Hornik K, Zeileis A. Unbiased recursive partitioning: A conditional inference framework. Journal of Computational and Graphical statistics. 2006 Sep 1;15(3):651-74.
Mogensen UB, Ishwaran H, Gerds TA. Evaluating random forests for survival analysis using prediction error curves. Journal of statistical software. 2012 Sep 18;50(11):1.
Periáñez Á, Saas A, Guitart A, Magne C. Churn prediction in mobile social games: Towards a complete assessment using survival ensembles. In2016 IEEE international conference on data science and advanced analytics (DSAA) 2016 Oct 17 (pp. 564-573). IEEE.
Rothenbuehler P, Runge J, Garcin F, Faltings B. Hidden markov models for churn prediction. In2015 sai intelligent systems conference (intellisys) 2015 Nov 10 (pp. 723-730). IEEE.
Runge J, Gao P, Garcin F, Faltings B. Churn prediction for high-value players in casual social games. In2014 IEEE conference on Computational Intelligence and Games 2014 Aug 26 (pp. 1-8). IEEE.
