“…In most churn prediction studies, session-related features, such as the number of logins, playtime, and the average time between games, were utilized as critical features that could indicate the extent to which players were engaged in a game [18] , [10] . Further studies added features of different kinds, such as social activities [25] , [31] , achievements [8] , and payments [51] , [52] . Most of the studies rely on machine learning-based approaches to deal with large and complex datasets.…”