Classification and Prediction of the Gaming Activity States in Online- Games Based on the Regime Switching Models

Abstract

The goal of the study is to construct an algorithm of determining the overall gaming activity state in online-games. Apart from the classification of states, the problem of predicting future gaming activity states has arisen. Both goals are accomplished by using multivariate econometric models with heterogeneous structure and the assumption of the hidden Markov dependency of the classes of states. In particular, Markov-Switching Vector Autoregression Models (MSVAR) have been used. The constructed theoretical methods have been applied to the measuring gaming activity in the famous online-game: "World of Tanks".