Machine Learning and Big Data Analytics at Tencent Games

• Activity Number: 602 - Game Analytics: How Data Science Transforms the Game Industry
• Type: Topic Contributed
• Date/Time: Thursday, August 1, 2019 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistical Learning and Data Science
• Author(s): Qiaolin Chen* and Xu Cheng and Jiachun Du and Botao Li and Zeng Zhao
• Companies: and Tencent and Tencent and Tencent and Tencent
• Keywords: Game Analytics; Machine Learning; Precision Marketing; Data Mining; Knowledge Graph; Complex Network

Abstract

At Tencent Games, all decisions are data driven. Data mining techniques are applied to huge data sets collected from 200+ games to provide end-to-end big data analytic solutions. This presentation explains (1) how we develop a scalable and comprehensive user profile system for billions of game users based on demographics, game activity, monetization, and social network, and (2) how we apply various predictive modeling techniques throughout customer life-cycle to improve game user experience and optimize targeted marketing. The different business scenarios we covered include new user acquisition, user monetization, user retention, item recommendation in game mall, and churn prediction. By operating based on the user segmentation, we successfully shape business and technical strategies and maximize income with lower costs.