Silicon Studio has created a mathematical model that determines how soon a user will get bored in a mobile game.
The algorithm predicts the day, the level, and even the reason why the gamer will stop entering the game.
We determine with a high degree of confidence at which level the user will stop, and how many days it will take him to reach it.
head of the game data research department, Silicon Studio.
The program uses the so-called “ensemble” method: a model that is based on many learning algorithms instead of one. Thus, the accuracy of the forecast becomes higher, correlations and alternative models are taken into account.
Every time we run a model, we use 1000 different submodels, each of which focuses on different variables and has different initial conditions.
Head of Game Data Research, Silicon Studio.
Within each submodel, the team also used a survival analysis algorithm.
These models are used in medical research and biology to know how individual cells will behave in the body.
Researchers have combined the power of survival algorithms and “ensemble” models in the field of video games for the first time.
Perianes added that the main priority in their research is to try to “prolong the life” of the game and increase its sales. To do this, it is important to understand the needs of users in order to create more exciting and stimulating games.
According to the scientist, developers can get this information from user data that they themselves store: purchase histories, contacts and actions.
Source: Money Control