Free-to-play mobile games monetize players through different business models, with higher player engagement leading to revenue increases. Consequently, the foremost goal of game designers and developers is to keep their audience engaged with the game for as long as possible. Studying and modeling player churn is, therefore, of the highest importance for game providers in this genre. This paper presents machine learning-based models for predicting player churn in a free-to-play mobile game. The dataset on which the research is based is collected in cooperation with a European game developer and comprises over four years of player records of a game belonging to the multiple-choice storytelling genre. Our initial analysis shows that user churn is a very significant problem, with a large portion of the players engaging with the game only briefly, thus presenting a potentially huge revenue loss. Presented models for churn prediction are trained based on varying learning periods (1–7 days) to encompass both very short-term players and longer-term players. Further, the predicted churn periods vary from 1–7 days. Obtained results show accuracies varying from 66% to 95%, depending on the considered periods.

This paper presents the results of a subjective study aimed at assessing different locomotion (may also be referred to as navigation) techniques in virtual reality environments. We implemented a simple virtual environment in which participants were asked to assess the following four techniques using the HTC Vive platform: teleportation, controller movement, controller movement with tunneling vision, and human joystick. A total of 29 participants took part in the study and rated perceived level of immersion, physical discomfort, and overall Quality of Experience for all tested locomotion techniques. Moreover, participants indicated whether they needed a break after a specific test due to physical discomfort. The results indicate that teleportation yields overall highest QoE and lowest discomfort.