Intensity is a very important characteristic of human behavior. Knowing behavior intensity you can make predictions about behavior, which can be used in a number of areas: sociology, economics etc. We present a Bayesian belief network estimating the behavior intensity based on data about last episodes of behavior. We consider the work of this model on the data about posting activity in the social network Vkontakte. The intensity rate predicted by the model and the actual intensity of users posting are compared.