| Abstract: |
Modern video games record each and every action of their players, generating extremely rich datasets that—with the help of machine learning methods—can serve to explore social and consumer dynamics and to predict user behavior. By properly using these data, game developers can not only provide players with a unique and personalized game experience, but also increase their engagement. In this talk, first I will review some of the main applications of game data science, such as churn prediction (foreseeing which players are about to leave the game) or item recommendation. Then I will focus on the prediction of lifetime value, which is crucial to provide a customized player experience and optimize monetization, through different deep learning methods, including convolutional neural networks, long short-term memory networks and multi-layer perceptrons. Finally, I will discuss how to implement these techniques in an operational environment (with huge volumes of data, even in the order of petabytes) from a machine learning engineering and big data infrastructure point of view. |
