The Number One Educational Resource for the Game Industry
-
Overview:
-
This talk will introduce a new churn analysis framework for online games, which utilizes XAI to automatically mine the possible churn cases, provides actionable advice for further specific improvements, and tracks and evaluate the "post-treatment" effects for each adjustment. We will also present the valuable general principles for collecting user churn features of high quality. The whole framework consists of three stages including churn prediction modeling, explanations generation, and the intervention (evaluation included). Besides the module details, we will also give some practical examples of how to deploy our system into different types of games such as card games and MMORPGs. It is very encouraging that our system has been successfully adopted into multiple popular online games released by NetEase. Equipped by our decision support system for actionable churn analysis, these games can make swift and effective moves for keeping both the newcomers and the current active players.
Recommended
