In this session Ron Luhtanen and Petri Mertanen will tell how Annalect Finland is using machine learning in creating predictive models, which are explaining and predicting user behavior from web analytics data. By default quite many web analytics systems report behavioral data on an aggregate level. This is inconvenient for in depth analysis, because it limits the methods to only different regression models. During the session you will hear how browser and session level data is used with modeling algorithm. We will wrap up the session with benefits and outputs of the modeling.